AKAI GXC-715D

Initial Issues

• Oxidized fascia, or seemingly so from advert.
• No PLAY 
• Seized head assembly
• Seized reel brake system.
• Later, discovered playback distortion mainly in the left channel.
  
  

Regarding the fascia, the gamble purchase paid off, as almost all of the apparent oxidized marks were gently removed. Using firstly warm soapy water and rinsing, followed later by 'painting' Basso on the surface, then allowing it to dry, the marks came off.

Basic Re-capping of the PSU

Before any examination of the dysfunctional PLAY action was investigated, the PSU had to be re-capped.

'No PLAY' identified

The PLAY solenoid was inoperative, and the mechanism had also seized up - this had to be freed, cleaned, and lubricated. Also new PNP solenoid transistors for Tr32 and Tr37 were inserted, they were the BD140. All good.

For now, the schematic showing some of the fixes, and some identified areas of interest.

Transport Images

Task - replace old idler with new RW/FF and take-up spool idler. 

 

 

Observe the 'flat' on the old idler tyre. This gave rise to a rapid repetitive 'knock' sound  during RW and FF. The idler in the background was not used, but another I had in stock was.

 

The new idler is permanently pressed up against the RW/FF/Take-up knurled gear. A design flaw?

Initially many screws were difficult to turn - at this point it is very important that perfect fitting JIS or Philips screw drivers are used. The reason is perhaps not obvious, but experienced technicians will tell you it is very easy to distort the head before the screw begins to turn. The area around the neck of each 'difficult' screw was soaked in acetone and cleaned, then later a warm freeing oil was applied, and allowed to 'work' its magic! I think the latter was the most effective treatment.

DC Motor Loading

Using the F3 fuse holder, the AC motor circuit current was measured. During a free-running period, IF3 ~ 31mA, and under PLAY conditions IF3 ~ 34mA. These are exceptionally low current figures, meaning that the effective DC current at the brush-commutator interface is even lower - excellent!  These figures translate into very low torque loading on the motor.

The motor pulley is made from aluminium and a 'press fit' on to the shaft, and so no attempt was made to remove it. With the motor out of circuit, it was powered separately, lubricated at the neck where the bearing resides, and the pulley was cleaned by gentle working of 400 grit paper to its profile. All residual marks were removed. All good.

PSU Measurements

With TR41 and TR42 being voltage regulator transistors, I thought it prudent to measure the DC emitter current under various load conditions for both. The reason being that the heat sink does get quite hot during long periods of PLAY and PLAY+RECORD. 

Emitter current readings IE were more convenient to make than collector current IC, and since IC ~ IE, we can exploit that convenience.

(a) TR41: (VE to Earth ~ 5.5v) VCE ~ 9.5v - 5.5v

Supplying: M54410P Logic, and lighting for commands.

No PLAY/REC/RW/FF: IE ~ 118mA. Command lighting uses 0mA.

PLAY:  IE ~ 240mA, includes PLAY lighting of approx 128mA

RECORD + PLAY: IE ~ 370mA, includes PLAY+REC lighting of ~ 260mA

RW/FF: IE ~ 240mA, includes RW/FF lighting of approx 135mA

Examining the TR41 transistor power consumption, and considering its highest working conditions, we have - 

P = VCE x IC = (9.5v-5.5v) x 0.370A, or approximately 1.5 watts. 

Both TR41 and TR42 are now BD139 NPN low power transistors, which have a rating of about 12.5 watts. They are both 'connected' to the original heatsink with new mica sheets.

 

(b) TR42: (VE wrt to Earth ~ 21.5v) VCE ~ 33v - 21.5v

Supplying: Audio record and playback sections, autostop, and bias oscillator. Command switch circuits.

No PLAY/REC/RW/FF: IE ~ 133mA

PLAY: IE ~ 137mA.

PLAY + REC: IE ~ 168mA. The bias oscillator now running.

P = VCE x IC ~ (33v-21.5v) x 0.168A, or approx 1.9 watts.

Although the power consumption isn't high, the heatsink does get quite hot. This wouldn't be an issue normally, however, the heatsink is close to the transport chamber. Disengaging the line to supply the command (PAUSE/REC/RW/PLAY/FF) lamps significantly reduces heat in that area, and so an LED based lighting solution will be sought in the future.

----------

Unregulated Voltage Supplies

Collector of TR42, unregulated (33v..32v) supplies PAUSE solenoid.

Collector of TR41, unregulated approximately 9.5v supplies Reel Motor and PLAY/REC solenoid logic only.

PLAY, RECORD, and brake solenoids are supplied from the 'D3' labelled full wave rectifier, and C8.

This article on the restoration of the AKAI GXC-715D remains unfinished, therefore expect alterations, corrections, and additions.

03/08/2026

 

AKAI GX-F31 Cassette Deck

 AKAI GX-F31

The AKAI GX-F31 with 
a wood effect vinyl wrap.

Distortion of Output at Line Out and Headphone.

 

After playing the deck for a short while, distortion began to 'creep' into the left channel and also began to fade in volume. 

 

The fault was difficult to pin point so as a basic starting base I re-capped the PSU (Power Supply Unit) and later re-capped most of the area surrounding the Dolby IC. However, this made no difference.

 

On injecting a test tone of 1Khz at line in, the muting at the source terminal of JFET 2SK177 was good, but not at this JFET's drain side for the left channel - there was no signal here for audio line out.

 

Much later, it was only 'by chance' I checked various DC voltages around the Dolby circuit and the sound mute area that I discovered that two electrolytic capacitors C97 and C97b were in opposite polarity to that suggested by printed circuit board markings. 

 

These voltages across 'C114' (according to the service manual) where in fact marked as C97 on the actual board were approximately -6v. This is the correct voltage, but as stated above the original electrolytic capacitors were placed incorrectly.  

 

 

 

Over a period of time 'C114' must have begun to break down and the internal electrolyte deteriorated to such an extent that the capacitor was partially conducting?

 

In the mean time, prior to knowing what the fault was, both JFETs Tr36 (2SK117) were removed and the source and drain connections were shorted. This allowed the AKAI GX-F31's audio to work, but at the cost of no muting.

 

Once the fault was rationalised, and new caps were put into circuit at opposite polarity to that printed on the PCB at C97 and C97b, a search for a suitable JFET had begun. 

 

In my stocks I have some 2SK30A, 2N5457, J113, and other N-channel JFETs. 

 

For the muting to work, the Gate to Source Vgs(OFF) voltage will have to be negatively high enough to invoke n-channel depletion, or a 'pinch off' state of the n-channel semiconductor.

 

A temporary three-pin 'platform' or 3-pin 'mount' was soldered in so various JFETs could be tested without resorting to soldering and de-soldering on the main board. A 2N5457 tested fine, but so too did the J113. The latter is not normally used for audio applications, but it's job in this context would be to - 

 

(a) act as a small resistance path during PLAY, REC/PLAY, or 

 

(b) to act to impede signals (very high resistance) to line out during the depleted n-channel or Vgs(OFF) state.

 

 

Note: the original pin-out configuration from this view is: Drain, Gate, Source, but the J113 pinout is Drain, Source, Gate, so the leads had to be carefully bent to fit the configuration.

 

The J113 JFET solution works flawlessly.

 

Muting Circuit Analysis

 

This line out muting circuit is unusual, and rather complicated. I wondered why AKAI adopted this design?, perhaps it was 'just to be different' and so they (like others) could claim this to be another new design? 

 

There are two switching voltage lines -

 

(a) At '-7.3v' or actually measured as -7.44v permanently, and ...

 

(b) the other line switches between +26.2v (Mute=ON) and -3.2v (Mute=OFF). 

 

Here and below, these are represented by a single equivalent voltage source for simplicity. It is assumed their Thevenin equivalent impedance is low.

 

 

 

Line out MUTED

 

This occurs when the GX-F31 is not in PLAY, and REC+PLAY modes.

 

The RL resistance is an additional and there merely for completeness, it plays little part in the muting analysis. 

 

An assumption was made, and that was - the transistor's base-emitter junction when forward biased exhibits an approximate dc resistance of 1K ohm, perhaps more typically up to 1.5 K ohms. 

Confusion must not be made with dynamic (small signal) resistance 1/hoe. 

Line Out Not MUTED

This is a hand written approximation, no simulation software used, although later the figures were verified by measurements. 

 

 23-06-2026

 

The content of this article

may be modified or corrected  without notice  

 

 

ITT 8025 Cassette Deck

 ITT 8025 Cassette Deck

Some pictures and commentary while I've been working on it for a future sale.

Cassette Transport

Rightmost is the back tension brake and 'new' felt pad.

Drive Belt: Out of sight is a 79mm x 5mm flat belt for the motor and two flywheels.

Lapped Record Head

Original Playback Head Lapped

Some New E-clips

 

There is a small issue with this transport, namely head assembly penetration, and a slight leftward misalignment of the record head. 

Observe that the record head is aligned slightly to the left, although it does not catch any cassette shell to date.

 

 
With record head removed.

With record, playback head, and head assembly plate removed.

Head plate, was later partially reamed to allow a slightly further drop or effective reduction in head penetration during PLAY.

Three bolts hold the returned head plate. The threads for the bolts are located underneath the head plate. The effective drop in head penetration during PLAY is now approximately 0.5mm less than previous.

All heads returned and re-aligned using a M-300 gauge (for tilt), a Hans Peter Roth 'narrow track' 1Khz head height tape, and an ABEX 10Khz full track reference were used for playback head alignment.

However, there was another yet-to-be-fixed issue. And that was - the record head was still too far left-of-centre, so the whole process was repeated above, but this time extra reaming to allow the plate to be shift a little more to the left. 

 

The result was to ensure that the small record head could penetrate (almost) dead-center into the cassette shell aperture.

 

Recording Failure

 

Much later a new issue surfaced - the ITT 8025 was only recording in the left channel.

 

Checking the bias signal at pin 1 of the HA-1457 pre-emphasis voltage amplifier revealed that one of the two HA-1457 amps had probably failed. 

 

During a recording session, pin 1 normally comprises of the (bias-trap attenuated) 85Khz bias signal and the audio. The overall display resembles Amplitude Modulation with the carrier 'amplitude modulated' with a test signal, although technically it is not. 

 

A high frequency test tone is best employed to observe the two on an oscilloscope since pre-emphasis ('record EQ') exaggerates the tones above 400Hz. However, an audio 10Khz test tone can only be seen if sufficiently large, but was found only in the left channel; the right channel was dead.

My assumption of a faulty HA1457 later turned out to be incorrect, however that didn't stop me from removing the old HA1457 ICs and replacing them with a TL071 OP Amp based solution with step-down voltage 1N4148 diodes in series.

 

Original HA 1457 Pre-emphasis Amplifier

HA 1457 Replaced with a TL071 Circuit Solution

 
It wasn't until I had ran the unit with the new pre-emphasis pre-amps that I realized the fault may have been caused by a faulty Dolby integrated circuit, namely the LM-1011 in the recording chain.

After interrogating input pins 5 and output pins 7 on each LM-1011, the right channel had clearly failed at the output on pin 7. Both ICs were carefully removed, and with new 16 pin DIL sockets in exchange, I was tempted to try the  Signetics NE545B Dolby chip.

After consulting the datasheets on the NE545B series, I was partially confident these would work. 

The supply voltage to the previous LM-1011 was 20v (exact), and the maximum rating of the NE545B is '24v' or is it 20v? So, just be to be 'on the safe side', I wired into the supply rail a single 1N4148 signal diode to drop the voltage by about 0.7v. This was done by cutting the track on the solder side of the board and positioning just one 1N4148 signal diode.

Two Dual In Line ('DIL') Sockets host the NE545B Dolby ICs.

 

Both left and right channels are now recording, and playing back well.

{More on this may follow} 

Tape Transport and Tape-Slowing 

Playing some cassette shells still resulted in the tape slowing down, which I initially thought was due to the record head being too left-sided, and overall head penetration too excessive?

Well, the above were indeed issues to be corrected if possible, but the nagging issue of some cassette tapes and their shells causing mischief for unrestricted transportation still proved a stumbling point.

That is, until I notice something in an earlier image I took of the deck. It became apparent that there was a conflict between the supply tape guide and the cassette shell - the two sides were making contact on some cassette shells, most noticeably on an old 1970s AGFA, and 1990s TDK D46.

I realized much earlier that manually lifting the cassette shell during PLAY resolved the issue, but I was still certain that the left-leaning record head and general head penetration were solely to blame. 

As a final attempt, I decided to check out the supply tape guide, its positioning and the baseline on which the shell sits. After loosening the baseline guides, I realized that these had probably slipped downwards over the years through casual insertions of the tapes. And so it proved to be, by resetting the baseline height the whole transport became fully cooperative!

No more tape dragging! 

   

Tape Creasing:

How much the tape dragging problem concealed tape creasing I'll never know, but the machine does crease tape on some cassette shells - it seems to begin on the supply side. 

The old supply pinch roller measured 9mm x 8mm x 1.5mm, but I only had a single 9mm x 7mm x 1.5mm in stock. This new roller worked, but 'end play' had to be reduced. End-play was reduced with suitable nylon washers. 

However the new roller did not eliminate any tape creasing. 

Tape creasing is sometimes caused by ...

(1) the tape being fed into the supply capstan/pinch side which is slightly uneven - either due to ineffective back tension, a poor cassette shell, or an out-of-parallel capstan/pinch roller line.

(2) on some occasions concurrently, it can also be invoked by insufficient supply capstan/pinch pressure, as the take-up side is perpetually pulling the tape along - the tape also 'micro' slips here?

(3) Head tilt which is not 90° to the plane of the traveling tape. There is obviously a margin of error here which would be acceptable. 

For this ITT 8025, increasing back tension will stop tape creasing, but on this machine back tension was (I believe) sufficient. Increasing capstan to pinch roller pressure also cured this problem, and to achieve this I had to configure the tension springs into a slightly more tensile state.

I first swapped the springs - I may have made a mistake earlier by swapping the, I cannot remember?

Secondly, the tension of each tension spring was increased a little more by placing a plastic sheath over the anchor point in each case as illustrated below.  

Finally, the playback head tilt was checked again and altered very slightly to favour an even tape wrap across the head.  An M-300 gauge is essential.

 

Tape creasing tests were performed on cassettes without a tape pad.

 

Reverting to Original Rollers

After some thinking, I later decided to return the original rollers - both supply and take-up rollers were in very good condition. Of course - no tape creasing, and tape flow is very stable.

 

 

Wow & Flutter:

Just a quick test: 0.07% ... 0.08% WRMS so far. 

Frequency Response: 

At -20dB, casual white noise 60 second recorded test - better than 30Hz to 17,000Hz using a TDK D46.

More may follow.

(20/04/2026)   

Sold, May 2026.

 

 

NEAL 102: Take-up Spool Torque

 NEAL 102: Take-up Spool Torque

 

 

The idler that drives the take-up spool's torque needed attention since the acting torque was either weak or intermittemt. 

 

Initially, the edge of the rubber idler was 'roughed up' and cleaned, and treated with Rubber Renue, but a few hours into PLAY, the same problem arose - torque was intermittent, and this could be heard on playback.

 

The solution was to either replace the idler, or remove it completely and re-work its surface or edge.

 

Flywheel

Audio Board & Removed Flywheel

Flywheel Removed, Idler Exposed

Idler Mechanism Removed

Seperated Components

Note the idler's form, which is unlike any idler normally made.

The rubber idler was first carefully sanded on its perimeter, cleaned, then soaked in Rubber Renue for about 5 minutes. After wiping down and allowing to dry, the idler was returned to its mechanism.

 

 

Out of sight and on the other side, there is a tension spring which needs to be returned, this provides the necessary force so that the idler presses between flywheel and take-up spool gear.

 

Idler Spring Returned.

JFET Measurements

JFET Measurements

 

As measured by the PEAK DCA75 Pro: 12mA current limitation. 

Note: JFETs do vary from their specification from sample to sample. On offer here is just a guide, which may be cross-referenced with official datasheets. 

(1) OnSemi: J112

Symmetrical Source/Drain

Vgs(off)=-3.66V at Id=5.2µA

Vgs(on)=-2.73V at Id=5.01mA

gfs=9.8mA/V at Id=3.0mA to 5.0mA

Idss>12.00mA at Vds=0.47V

Rds(on)=24.4Ω at Id=5.0mA and Vgs=0.0V

(2) OnSemi: J113

Symmetrical Source/Drain

Vgs(off)=-1.82V at Id=5.3µA

Vgs(on)=-1.00V at Id=5.00mA

gfs=11.0mA/V at Id=3.0mA to 5.0mA

Idss>12.00mA at Vds=0.82V

Rds(on)=34.3Ω at Id=5.0mA and Vgs=0.0V

(3) N-Channel: 2N5457 JFET

Symmetrical Source/Drain

Vgs(off)=-0.35V at Id=4.8µA

Vgs(on)=0.17V at Id=5.00mA

gfs=21.9mA/V at Id=3.0mA to 5.0mA

Idss=1.74mA at Vds=3.01V

Rds(on)=27.6Ω at Id=5.0mA and Vgs=0.0V

(4) N-Channel JFET J109

Symmetrical Source/Drain

Vgs(off)=-3.59V at Id=4.7µA

Vgs(on)=-3.01V at Id=5.00mA

gfs=26.9mA/V at Id=3.0mA to 5.0mA

Idss>12.00mA at Vds=0.09V

Rds(on)=4.4Ω at Id=5.0mA and Vgs=0.0V