Monday, 30 August 2021

Sony TC-138SD

 Sony TC-138SD Cassette Deck (1975) Restoration/Servicing

I half promised myself not to buy another cassette deck, and although I do tend to 'window shop' on ebay, many machines I view, I eventually pass without a second thought. Then, sometime ago I saw the TC-138SD, which intrigued me since it is similar to the TC-136SD, but appeared to be of a much better build? According to the 1976 HiFi Year Book which I have owned since about 1977, these machines weigh about 7Kg, so it was a lot heavier than the TC-136SD. It also has a syncronized ac induction motor, unlike the 138SD.

There was one 138SD that caught my attention which got perpetually re-advertised and was finally down in price to £25 + £15 postage. I saw that the seller lived only 5Km from me, and so I contacted him and asked if he would sell it for £25 + £5 petrol money for delivering it to my door. And sure enough a few days later - I had another cassette deck!

Original Advert

Originally sold for - spares or repair, untested.


The cassette door had broken, but luckily,
the missing piece was inside the deck!


TC-138SD Restoration/Servicing

Within 20 minutes I had the machine basically working. I very first thing I did was to switch-clean (Servisol Super 10) all contacts and switches, paying special attention to the multi-functional switch that facilitates playback, and record modes. This is important because many adverts show their machines with one or more of the VU meters oscillating irrespective of mode. Over the years dirt, moisture, and low level oxidation - hence corrosion takes place and will 'bridge' circuit paths.


After lengthy and detailed service
work on the Sony TC-138SD, I now have a fully working machine. I've listed to date, the work done on the TC-138SD  -

  • Deep clean of all switches and contacts using Servisol Super 10
    (not WD 40 or equivalent!)
  • Clean all internal areas and mechanisms.
  • Lubricate the capstan
  • Clean the heads, demagnetise capstan and the heads.
  • Clean the pinch roller with diluted acetone, and later Rubber Renue.
  • Fully re-cap all audio board electrolytic capacitors
  • Replaced original motor run capacitor: now 1.5uF 400v rating.
  • Fully replace all audio board transistors with suitable equivalents,
    mainly C1815-GR, and C1845FTA NPN transistors.
  • Add a bias (105Khz 'carrier' amplitude) circuit modification -
    allows the user to alter bias continuously by a larger margin than originally.
  • Add a bias trap circuit modification for variable adjustment -
    better or worse rejection? Testing to follow at a later date.

Initially, the deck was in good condition, just a little dusty and dirty.

Playback Equilisation Adjustment

The EQ playback control (playback frequency curve, called de-emphasis) is integrated into the feedback loop of the pre-amplifier circuit. I strongly suspected the playback EQ response did not conform to the later IEC standard adopted and modified in the early 1980s - the IEC Prague 1981 standard? Some tape enthusiasts call this 1981 standard 'IEC 2', so I'll refer to it as IEC 2.

The playback EQ for Normal and Ferro Chrome/Chromium Dioxide tapes together with playback level (PB Level) conrol are accessed at the front of the deck.

To give more flexibility and higher HF boost,
I later exchanged the 2kΩ potentiometer for a 10k,
although 5k
Ω would have been sufficient.

Playback Calibration 

This was always going to be an issue - matching the playback frequency response of the TC-138SD with an ideal IEC Prague 1981 'IEC 2' calibration playback tape. 

Ideally, on playback from an IEC Prague 1981 tape, we should observe a flat response somewhere around 333Hz to 10Khz.

I am suggesting that pre-1981 machines (apart from Nakamichi?) generally adhere to an older pre-emphasis/de-emphasis standard, which at this moment, I am unfamiliar with. Is it a revised IEC/RIAA curve for 4.75cm/s? - I don't know. It seems that the general consensus within Tapehead forums is that manufacturers broadly abided to the old (pre-1981) IEC standard, but tweaked responses to match the ability of their record/playback heads having to make a balance between frequency response, noise, and distortion.

The Sony machines that I have seem to opt for more aggressive HF cut in their de-emphasis?, the Nakamichi - less so?

My solution, although not ideal was set some kind of compromise. The TC-138SD will be used as a recording machine, but mainly as a playback machine.

A Nakamichi IEC 2 tape? 

Some time ago I made a Type I (TDK D C-46) frequency sweep tape at about -10dB (ref: dolby level) on my Nakamichi DR10. I used Audacity software to create a series of 100s of sweeps between 333Hz to 10,000Hz. This was transfered to an iPod and then used as a signal source for the Nakamichi DR10. The recording/playback response of the DR10 was flat from 333Hz to 10,000Hz. This flatness was also confirmed on my Revox B215, ±1dB at worst, and within ±1.5dB on a 1989 SD-35 Marantz deck.

Playback on the TC-138SD

Playing back the NAK DR10 (a now 'IEC 2' equivalent?) tape on the TC-138SD allowed me to observe the playback response of the old Sony. I could easily set the PB so that 333Hz and 10Khz were equal in amplitude, but at a cost - a large mid-band rise in output of about 3dB to 4dB somewhere between 1Khz and 5Khz. 

After much experiment and thought, I decided on a playback drop of -3.5dB to -4dB at 10Khz from the NAK DR10 ('IEC 2' equivalent?) self-made test tape. This significantly reduced the mid-band 'hump' to +1.2dB/+2.6dB. I could have reduced the R-ch 'hump' to 1.2dB but this caused a further drop in response at 10Khz. As can be seen, there is a lack of symmetry here.

Bias settings were tweaked to give the TC-138SD a record/playback difference at 333Hz/10Khz of about -1dB, that is 10Khz is down by -1dB with reference to 333Hz. This of course gave a rise of about +3dB at 10Khz if played back on the 'IEC 2' spec Nak DR10, or Revox B215. 

I may at a later date modify the record EQ to dampen any further mid-upper band rise in frequency before dropping to -1dB at 10Khz as stated above.

Since the TC-138SD has still, but now a small rise in output between 1Khz and 5Khz, I deliberately reduced the internal record level calibration back by about -2.5dB at 333Hz for post 1990 TDK D60 tapes. This ensured the meters on playback won't jump into the red zone too often. 

Recordings made with the TC-138SD genrally sound good, balanced, and match up well on the meters of other old pre-1981 Sony, and post 1981 machines. It's not ideal, but it works and the mismatch isn't obvious.

High Frequency Saturation

If I do have one criticism at this moment, it is that recorded high frequency distortion or saturation can be heard on some tapes when listening back to speech sibilances, ie the high frequency transients in speech. Even my TC-134SD doesn't exhibit such record/playback issues, and yet uses the same record/replay head!? There is no HF distortion/saturation issues for the 138SD on playback from another 'better' machine. I need to look into this matter.

Recording Bias, and Bias Trap Control 

Recording Bias

Once again, there is going to be some discrepancy between the ideal IEC-Prague 1981 'IEC 2' recording EQ (pre-emphasis) and the recording EQ ('IEC 1'?) curve adopted by Sony back in the 1970s. I do have some control over this, but for now - let's see where I can go without modifying the recording pre-emphasis.

I wanted better control over the bias - effectively, this is a high frequency 105Khz bias signal 'carrier'. 

In brief - controlling the bias, controls the amplitude of the 105Khz bias signal. 

Bias adjustment also influences HF recording levels, output levels, noise, and distortion. The trick is - to get the balance about right.

The audio then 'sits' on this bias 'carrier' as both signals are used to produce a varying magnetic field at the head face gap.

I removed the original 8pF trimmer, and desoldered the capacitance bridges from the capacitance bank. A new 10~60pF trimmer (with high voltage rating) was inserted into the same place and no capacitor bridge was re-soldered.

I am now able to modify the 105Khz bias amplitude at the head from about 15v (or less?) to about 46v peak.

Bias Trap

The purpose of the 'trap' is to block the bias high frequency signal, and more importantly - the threatening high voltage that can leak back into the audio section. The trap comprises of a simple bandstop resonance circuit, which at resonance (about 105Khz for the 138SD) acts as a very high impedance to the bias signal, but not the audio. 

In theory the rejection at resonance is infinite, but in reality due to inductor resistances (typically 10Ω - 30Ω?) and capacitance leakages/losses, the rejection is finite. 

Oscilloscope traces show that on the record head side of the bias trap (bandstop) LC filter, the 105Khz voltage is about 35v peak, and on the other side of the LC filter - less than 0.5v/0.7v (L/R) peak. I would have liked to have seen a better rejection - I neglected to compare the original rejection capabilities.

As a comparison, I later measured bias frequency amplitudes either side of the bandstop filter on a Sony TC-K45 deck - they were 15v and 0.36v.

So comparing this TC-138SD's 35v:0.6v (average), to the TC-K45's 15v:0.36v, it can be seen that the two ratios compare well, that is - rejection is similar.

The intention of adding a bias trap trimming capacitor was to tweak for best resonance and so best rejection. Oscilloscope signal observation allowed me to minimise this bias leakage.

Bias Trap Modification: I eventually replaced the default 91pF for two high voltage rated capacitors in parallel. They are a 68pF 1KV rated, and a 10~60pF 250v rated trimmer. Before soldering into circuit I measured both so that in the parallel configuration they became ~ 90pF. Turning the trimmer allowed me to observe minimum leakage voltage fed back to the audio section.

Bias Trap - later revisions: Later, I may experiment by inserting a lower variable inductance and a higher value capacitor. Better bias signal rejection may be obtained from lower resistance in the inductor choke coil?

 


Access to -
Meter Calibration (out of shot at rear),
Record Levels (as indicated),
Bias, and Bias Trap access is easier than earlier Sony models.


Small 10pF .. 60pF (high voltage rated)
trimmer capacitors were adapted to ...
(i) add variable bias control,
(ii) add variable, and better bias trap rejection?

The Working Sony TC-138SD

The cassette lid will be repaired and fitted later.




07/09/2021. Finished

This blog/article is still subject to corrections or minor modifications if the need arises.


Monday, 14 June 2021

Revox A77 Servicing

 The Revox A77 MKIII

 

This blog/article is a work on the progress of servicing my 1971 Revox A77. It will be periodically updated, amended, and corrected until it is finished. When finished, this paragraph will be erased. (02/07/2021)

A factory stamped 'November 1971' Revox A77 that's been in the family since 1972.

The same date was also originally stamped on the heads.

This blog/article is intended to be a servicing account, let's see how far I can go?

Last time I switched this on was around 2003-2005. Before switching the machine on though, I checked that all three motors (Reels, and the capstan) were turning smoothly, other internal mechanisms ran/moved smoothly, contact and switched cleaned here and there, and finally changed some potentially dangerous capacitors that are known for catching fire.

14/06/2021: Report so far - it's working!

Playing original AGFA PE36 Tape.





Inside rear view - before any capacitor replacement.


Initial Inspections
 
Headwear and Guides
 
Record/Replay Head wear: the flat spots are quite wide.
Tape Lifter: small amount of wear, but may need replacing.
Erase Head: No signs of wear
Left Tape Guide: Runs smoothly
Pinch Roller: Free running and cleaned with Rubber Renue *
Capstan Shaft: Cleaned *
(*after this picture was taken)

 

 

 

 
Reel Motor and Spindles
 
I checked the free movement of the reel motor spindles - they spin easily with no apparent excessive internal friction. The shafts will be cleaned and lubricated later.
 


Some Initial Capacitor Replacements

Tape Drive Board: I initially replaced the said board's capacitors with metalised polyester 0.47uF 250v dc units, but later changed again to 0.47Uf 275 vac 'safety capacitors' in yellow. I'm not sure if there's any benefit to this, but it seems to be the general consensus?
 
Actual units in the Revox A77 now in yellow.

Originally -


Initial replacements -



Power Supply & Regulator Board.

The old original 2500uF canister type electrolytic capacitor has been replaced by a 3300uF 50v rated device - shown at the rear. A blue radial JB type is a 1000uF direct replacement for the old 1000uF axial type. 
 
I hope later to replace the '2500uF' original with a KEMET 100V 3300UF ALS31A332DB100. This has a high ripple current rating. 
 
An old 3.3uF tantalum (not shown) in the regulator circuit side has also been replaced, but by a standard aluminium electrolytic. All replacements have higher voltage ratings than their older counterparts.

There are two dc power supplies on the A77 -
 
(1) A 27v dc circuit containing the 3300uF capacitor is the only form of 'regulation' and/or smoothing which serves the transport relays, solenoids, power and meter lamps.

(2) A 21v dc regulated circuit where the 1000uF capacitor resides and serves its purpose for  - audio electronics, capstan motor speed control, and the end-of-tape lamp.


These are the initial and probably necessary replacements before I could be reasonably confident of working the deck. There are no running hot issues at all.
 
Getting the Revox A77 up and running has given me a good measure of confidence to playback some early mono 1960s recordings made by my father that I wanted to digitise.
 
So far the Revox A77 is stable, and generally working well.
 
I'm aware that there are several minor issues to address -
 
  • Complete the re-capping of all electrolytic capacitors, including tantalum types.
  • New Motor Run Capacitors. (Done: 01/07/2021)
  • End of tape switch off issue
  • Tape guide bearings may be improved by replacing?
  • Reel brakes cleaned, better than before - re-clean?
  • Tape lifter may need replacing or modifying?
  • Reel motors may need some lubrication?
  • Capstan motor may need lubrication?
  • Bias trap band-stop filter refinement? (Done: June 2021)
  • Bias amplitude trimming to obtain a frequency response 30Hz .. 16Khz ±3dB at 9.5cm/sec.
  • New Record and Replay 1/2 track heads? - will need to create an azimuth tape, and calibrate playback levels from a known reference tape.

Bias Traps

A 120Khz high voltage 'bias' frequency generator on board the A77 is a neccessary building block to successful analogue recording on tape. A bias 'carrier' voltage is applied to the record heads, and the recorded signal (music. speech etc) 'sits' on top of this 120Khz bias signal. This modulating scheme is similar to Amplitude Modulation 'AM", although it is not mathematically the same. The result of applying a bias carrier is a linear, low distortion magnetic field image of the audio transfered to tape.

However, some parts of the overall A77 circuits need to be isolated from this 120Khz bias carrier, and so Bias Traps are employed - in particular, the recording, and playback amplifiers.

 

Underside of Revox A77
The Bias Traps for Record, and Playback circuit
access points have been opened up ready for fine tuning.


The recording circuit schematic is shown below, and a test point is established for adjustment of the bias trap. The bias trap is a standard bandstop filter where at approximately 120Khz, the inductive and capacitive reactances are (theoretically) opposite and equal in phase. It is at this resonant frequency where the impedance of the LC circuit is theoretically infinite, however, in practice owing to small internal resistances and leakages this is never attained, but is sufficient enough to work well.

The Record Amplifier Schematic
I later traced and then marked on the underside of the board
a point where I could use an oscilloscope probe.


Trapping the Bias Signal: Record Circuit

Pressing Record and Play initiates the record circuits and the bias carrier is generated. Turning the small trimmer choke L501 (for CHI on card CHI, and CHII on card CHII) with an insulated screwdriver I observed the 120Khz signal, and turned for minimum amplitude - it never diminishes to zero. At this test point, the minimum amplitudes for both CHI and CHII were down to approximately 50-75mV peak.
 

Trapping the Bias Signal: Playback Circuit

After engaging Record and Play again the trimmer chokes were adjusted for the playback circuit. The test point this time is at line out, ie at the phono sockets. And again, we aim for minimum 120Khz bias leakage voltage. The bias signal does not completely diminish.
 


With Ampex 499 tape wound on to the Revox A77, setting record and play set to ON, and with a 1Khz test tone applied at the input set to 0VU, the recorded 'NAB' playback 1khz output at line out was approximately 6v peak to peak. 

Below shows the oscilloscope trace showing the best 120Khz bias bandstop attenuation results (without 1Khz modulation) -

 Note the crossover distortion - probably due to the non-linear B vs I  charactersitics of the choke at low levels of excitation current.

As can be seen, the amplitude is approximately 100mV peak to peak (~35mV rms) on CHI and 170mv peak to peak (~60mV rms) on CHII. The best figure equates to approximately 20Log(0.1/6) ~ -36dB lower than the audio signal at 0VU.

Both original 390pF bias-trap circuit capacitors were replacd by newer 1% tolerances versions. 

Revox A77 Motor Run Capacitors

Employed to produce neccessary current phase shifts so that all 3 motors experience the correct rotating magnetic field phasing.

Old motor run capacitors in the A77 are known to be suspect, especially after 20 years and more, so I need to change these. So far I've had no problem, but it's best to be safe than sorry.

Old A77 Motor Run Metalised Paper Capacitors
2 x 4.3uF, 1 x 3.5uF


Just ordered new Ducati caps ...


New Motor Run Capacitors Fitted

New A77 Motor Run Ducati Capacitors: 2 x 4.5uF, 1 x 3.5uF
(Note the small difference: 4.5uF compared to 4.3uF)



02/07/2021: As this is a blog/article which is unfinished, there will be alterations and corrections to the text in the coming weeks and months ...

Saturday, 24 April 2021

Sony TC-134SD (c1972) Repaired.

 Sony TC-134SD

About 3 weeks ago, I purchased this from a charity shop for £8.50 + £5.50 postage, but I donated an additional £15 as it was for a children's hospice.

After many hours of restoration work, the unit is now working well. 

However, there seems to be no internal circuitry dedicated for chromium dioxide tape record bias (bias frequency amplitude) or playback equalisation, despite the 'Cro2' setting? There are however, recording pre-emphasis circuits for both 'Normal' and 'Cro2' chromium dioxide tapes.

Quick Summary

Well, I've restored the machine to good working level - but not totally satisfactory.

A complete clean of all switches, a complete recapping of all electrolytic capacitors, some mylar capacitors, and full transistor replacement has been undertaken, including the dolby circuit. Even the bias oscillator transistor, resistor, and capacitors were replaced.

However, four issues still remain.

(1) Cro2: No apparent circuitry for satisfactory use of Cro2 tapes? While the Cro2 switch activates to shape the pre-emphasis for Cro2 use, there are no changes in the amplitude of the Cro2 bias frequency 'carrier'!? The bias frequency carrier for Cro2 tapes is usually higher, or was this standard practice back in 1972? Nor does the Cro2 switch activate an alternative playback EQ, ie Cro2 de-emphasis.

(2) Frequency Response: Poor playback, and record frequency response - less than quoted in the specification. After some considerable investigation I strongly suspect the Rec/Replay head is the problem?

(3) Low-level Record Noise in Left Channel: Very low-level, but intermittent additional white noise ('hiss') in the left channel after recording. It almost sounds like 'drop out' white noise. The source of this is still a mystery, but the recording audio stages are not to blame. Some tapes record a louder 'noise' than others. This noise is not present when recording with Cro2 or Type II tapes? This has raised general Normal/Cro2 questions - I have not yet studied the machine for general Cro2/Type II tape useage. So perhaps the answer is somewhere there?

Incidentally, this noise that I refer to is an additional, audible noise much like broken 'hiss' when Dolby is set to OFF. It could also be described as slightly quirky - almost as if it was radio frequency interference.

(4) Tape Transport Speed: A speed check revealed that the induction motor is running 5% too fast. Alternative motor pulleys with a shorter radius may not be available?

This is the current state of the TC-134SD.

27/04/2021. (See later 03/07/2021)

Higher than Expected Recording Bias Noise (08/05/2021)

After much investigation, but by all means not fully conclusive, the excessive intermittent bias noise as described above may be solely the result of this deck's record head being heavily under-biased!? Although, I'm not sure why?

As part of my investigation, I de-soldered the original bank of bias capacitors and replaced the circuit with two (high voltage) variable (10pF .. 60pF) trimmers. I am now able to vary the 83Khz bias frequency peak voltage between about 5v and 40v on entry to the record head.


Original Circuit.

 

Modified Bias Circuit (subject to revisions).

 

The old capacitor bank was de-soldered and removed.
Two high-voltage rated variable trimmers were soldered
in parallel so that their effect is to add,
ie, (10pF .. 60pF) + (10pF .. 60pF).

 

Bias Frequency Carrier at approximately 18v peak.
 

The Sony PF145-3602A head employed in this machine is also used in other Sony machines. Of note - their bias voltages are approximately 40v (rms?) - which suggests that the TC-134SD I have here is under-biasing on recording?

 

Theory Reminder

A typical tape head magnetisation curve is show below. The graphs show B vs H, where B is the magnetic field (flux) density, and H is the 'magnetising force'. H is a function of I, where I is the current.

Note: The non-linearity of the B vs I

Simplified Record Head Diagram.
B field (Webers/m^2)
H (Ampere Turns)


The record signal 'sits' on the Bias Frequency Carrier.

Is this low bias deliberate?

Well, I'm not sure, but it appears to be so, as the pre-emphasis circuit components are all correct as indicated in the service manual. Also, the bank of bias capacitors found in circuit are also as stated in the service manual. 

Playback Flatness: An internal playback EQ potentiometer can be adjusted to obtain a reasonably flat playback response to about 8KHz/10KHz. Playback frequency flatness was checked using both a Nakamichi DR10-made, and a Revox B215-made frequency sweep tape, from 333Hz .. 10Khz.

The Probable Source of Additional Intermittent Bias Noise?

Before I explain my theory as to why the deck experiences this added low-level interference here are some observations I made -

Decreasing Bias Amplitude: If I decrease the bias frequency amplitude down to 10v or lower - the excessive bias noise disappears, which I expected as only the erase head 83Khz high frequency erase signal is now responsible for tape hiss. 

Increasing Bias Amplitude: If I increase the bias frequency amplitude to over 30v (peak), again the distracting additional bias noise also begins to fade away - leaving only erase noise and an expected 'clean' bias noise from the record head to be present. However, there is a cost to raising the bias higher - the recorded treble 8Khz/10Khz levels diminish.

The intermittent excessive bias noise issue only persists for bias amplitudes of around 15v ... 36v, peak. However, not all Type I ('Normal') tapes exhibit the same excessive bias noise - but there a general problem here nonetheless.

The thought then crossed my mind that older 'Normal' cassette tapes generally require lower bias to work optimally - so perhaps the TC-134SD was set up this way?

  

Compromises

In the service manual '7.5Khz' (not 10Khz, or higher) is mentioned as a reference point in achieving correct bias calibration, that is - a flat response to 7.5Khz. Most 'newer' Sony decks make reference to the amplitudes of a test signal at 1Khz and 10Khz as a guide to correct bias calibration.

So I decided set the bias to mirror the frequency response expectations of this deck. I have now set the recording bias so that I can acheive 100Hz .. 8Khz  to within -3dB. This increasing of bias (but lowering of recording frequency response) has substantially lessened the distracting added recording bias noise.

Head Wear Consequences

Now, here's my theory - although the record head on this deck shows almost no visible wear, there is wear - it is extremely small, only under a x8 loupe can it be seen. 

Tape head wear impacts on the head's ability to record to the tape, and I am beginning to suspect that random microscopic air-gap regions between the head and tape now exist for his machine. This implies that even recording bias noise (not to be confused with erase head erasure white noise) is going to be affected - and why not? If a tiny air-gap is introduced into the magnetic circuit, then permeability will suffer, hence the strength of B field will suffer too.

Currently, I have set both left and right channels at the head to receive 40v peak bias at 83.3Khz. I was tempted to increase bias to a higher level, but the TC-134SD's treble response would have been unbearable.

The revised bias and oscillator board is shown -

 

Circuit Board Components and Small Revisions

Below shows a copy of the main circuit diagram with some alterations, including comments/references I made at the time.

 

A mixture of C1845, C1815, and C945 NPN transistors -
I didn't have sufficient amount of C1815s at the time, so C945s were used.
All replacements were made in pairs.

Line In Response at 10Khz

I noticed quite early on that the 134SD exibited high frequency drop of around 2dB/3dB at just 10Khz!? This was tracked down and found to occur before the signal entered the dolby circuit and indeed before the MPX filter. Intuition suggested that I should lower C110/C210 below the rated 220pF, and so I replaced this for 150pF. Now the dip in response is around 1dB.




Increasing Transistor Bias at Pre-emphasis Stage

Within the recording pre-emphasis circuit, I increased R137/R237 from 330KΩ to 300K. Reason? - to forward bias the transistor a little more, there was an issue with the cleanness of the treble. This action resolved it.


14/05/2021

Additional Images


 


Sony PF145-3602A Head

I hope one day to be able to replace this head for another similar, or failing that - take a chance and get this head re-lapped? I am intrigued by this additional and distracting bias noise. 

One other matter to resolve too, is to replace the pinch roller. I also suspect the pinch roller is not applying the same amount of tension at the edge of the tape, as left-channel oscilloscope traces seem to suggest? 

Final Bias Circuit (17/05/2021)

After much experimentation, I've decided to re-do the bias circuit and accept that this deck has limitations, and a probable minor problem with head wear, particularly in the left channel?

I've finally settled for a record/playback frequency response to 8khz at approximately -1dB, where at 10Khz the record/playback response can dip to about -4dB! Lowering bias achives slightly better results to about -2.5dB at 10Khz, although at the expense of the excessive bias noise as indicated above. 

The excessive intermittent bias noise varies from tape to tape, while TDK D (1990s versions) are very low and acceptable, earlier TDK D (1980s) present listening problems.

The revised bias circuit now allows me to alter bias frequency from about 15v to just over 40v, peak.


Tape Speed

One remaining issue which has not been resolved is the high tape speed on this TC-134SD machine. It is somewhere between 4% .. 5% too fast.

===================

Pinch Roller to Capstan Issues (22/05/2021)

Ever had problems such as chewed up tape? Well, you're not alone!
 
Changing my old pinch roller for a new one presented problems - more than I will write about here. However, it is worth observing one of the main causes of crimpled tape is down to a poor, or worn pinch roller, misalignment, or excessive tape tension cause by the take-up spool.
 
The ultimate test for uniform traction?
Observe tape flow stability before capstan-tape-pinch contact!
The most stable system can be observed when there
is no tape pad. Tape pad can and does increase stability.
 
The old pinch roller had to be replaced - it was uneven, although wasn't creating too many problems.
 
After extracting the old 2mm steel pin (had to drill it out!), I initially settle for this solution -
 
 
It worked well, but using a steel bolt with a thread acting as both support and guide for the pinch roller troubled me, so I decided to use a NOS (New Old Stock) 2mm steel pin in the existing assembly -
 
 
Sufficient Pinch Roller Force?
 
As a matter of interest to me, I compared the pinch roller forces on the capstans of several other cassette decks, and found many of them to be high - higher than I initially assumed!

Currently, this latest solution for my TC-134SD exerts plenty of pressure, tape flow into the capstan is very stable with a tape pad.
 
Stability with No Tape Pad
 
However, as a real test for stability, and without a tape pad - tape flow stability is lessened, and if I allow Play to continue beween 5 and 30 seconds (sometimes minutes), the tape may 'ride up' the capstan and crimple the tape. This is resolved by adding more force to the pinch roller - the tape then 'rides down' the capstan and tape flow is stable once again. The reason this happens in this example is simple - excessive tape tension! Any engineered reduction in tape tension will return the non-tape pad cassette transport to stability.

Pinch Roller & Tape Summary

If you have transport stability problems, check ...

(a) pinch roller/capstan general parallel alignment (usually okay)

(b) pinch roller wear - the roller has to be 'grippy', its cross section must be unchanging.

(c) Tape-up spool tension - is this excessive?

(d) Too much 'play' in the pinch roller's axis?

 

***********************

 

Subject to the correction of mistakes, and additions.

22/05/2021.

*********************

New Motor Run Capacitor Installed (06/07/2021)

The previous 1uF 250V motor run capacitor seemed to be fine, nevertheless I chanced a replacement rated at 1uF 450V ±5%, and the deck now runs about 2% fast, messured via an ABEX 3.15Khz test tape and frequency counter. Compared to previous measurements of 5% too fast - a much better result!

The Original Metalised Paper 1uF 250v ac rated Capacitor.

 

 
New Motor Run Capacitor
 
I can only assume the phases of the rotating magnetic fields for this induction motor were not complimenting each other, forcing the motor to rotate at about 5% too fast? Even at a 5% increase in speed, music and speech recorded on other machines sounded almost a semitone higher in pitch and at a correspondingly higher tempo.

06/07/2021: New Rewind counter belt fitted, very fiddly job!
 
Approx 15/07/2021: Chinese brass-centred pinch rollers now replaced by an original Sony pinch roller. No observable 'wow' on recordings. The Chinese roller (for some reason) would frictionalise, causing excessive 'wow' - not sure why?

Bias Traps (19/07/2021)
 
It is very difficult to completely eliminate the high frequency bias carrier leaking into the final stages of the record section. My attempt to minimise leakage completely was to replace the bandstop filter (C138/C238 100pF in parallel with L103/L203 33mH), with an effective variable capacitance. I chose a 75pF (250v/500v rated?) capacitor in parallel with a 10pF-60pF trimmer and was able to minimise leakage to about 145mV in one channel, and abouit 125mV rms in the other.
 
I cannot remember exactly what the previous leakage voltage was, but I think it was rather high at ~ 400mV rms? Also, as the inductors L103/L203 were not air-core types, crossover distortion was observed - very similar to that seen on the bias traps for the Revox A77.


The 'new' Bias Trap is shown.
Note: The new revised circuit for general bias is not shown here, see Modified Bias Circuit diagram above.


So far, so good!

 
 
Current 'To Do' List

  • New 25mm diameter x 1mm/1.2mm square section belt for the auto-stop feature.

  • New narrower flat section capstan-flywheel drive belt needed. The current belt is too wide - this excessively wide belt tends to shift to one side of the capstan flywheel, and as a result may effectively be driving the capstan above its designed revs/second?
 
Blog/article is subject to the correction of mistakes, and additions.

20/07/2021.