Vintage Audio
this section covers some restoration work that I did on my vintage HI-FI gear
Article #1 | 09:26 PM Wednesday 09/03/2022

AKAI GX-95: Introduction

I have this cassette deck since 2006 or so but I have never had the chance to enjoy it fully.

It was previously tampered with by somebody that somehow managed to misalign the preamplifier output level of the tape head left reproduce channel. And it had some other issues with the mechanical block, especially in the tape path. I ended up buying a complete low mileage mechanical block from eBay. I replaced it and finally I could listen to cassettes. Unfortunately, at the time I did the mechanical block swap, I didn't had the Vintage Audio site and I didn't document anything.

From the start, I had this idea of converting this machine to quartz locked tape speed control. More on this, later on.

For a few reasons, I specifically wanted the first generation GX-95 unit, and not the MK2. I excluded A&D Japan-only units from the start due to scarcity and price tag. In my opinion, the standard AKAI GX-95 unit is a good unit for everyday use, both in terms of recording and playback.

Enough talking and let's see what I can do to this machine in order to improve its characteristics.


The following articles are not to be treated as do-it-yourself tutorials on how to fix, restore, rebuild, or improve the unit in cause. This was not my initial intention. But you can consider this whole content as a general guideline, should you decide to launch into such an adventure.

The entire documentation is just a reflection of my work and I cannot be held responsible if you damage your unit, or even harm yourself in the process.

Article #2 | 10:42 PM Wednesday 09/03/2022

AKAI GX-95: Technical Data

This cassette deck has the following technical characteristics.

Track System4 track 2 channel stereo system
Tape Speed4.76 cm/sec
Wow and Flutter<0.025% WRMS
<0.04% (DIN 45500)
Frequency Response20 Hz - 19,000 Hz (±3dB) - Fe2O3 tape
20 Hz - 20,000 Hz (±3dB) - CrO2 tape
20 Hz - 21,000 Hz (±3dB) - Metal tape
Distortion<0.6% (1,000 Hz / 0 VU) - Metal tape
Signal to Noise RatioBetter than 59 dB (tape peak level)
Heads1 x LC-OFC Super GX record head
1 x LC-OFC Super GX playback head
1 x Ferrite erase head
Motors1 x direct-drive linear torque BSL motor for capstan drive
1 x DC motor for reel drive
1 x DC motor for mechanism operation
Power Voltage220 V at 50 Hz
Dimensions / Weight154 mm (H), 460 mm (W), 350 mm (D) / 10.2 kg

Article #3 | 07:31 PM Thursday 10/03/2022

AKAI GX-95: Parts List

I have assembled a partial parts list for this cassette deck.

Descriptions and Parts Listings

Let's start with a parts list for the quartz locked servo tape speed control circuit. This functionality is not part of the original implementation and nor is it specified in the service manual. The parts list was obtained from reverse engineering with the help of a member of the Tapeheads community that added some pictures of the SYSCON PCB assembly from an AKAI GX-Z9100EV unit. Thank you for that!

The quartz locked control circuit is located on the System Controller (SYSCON) printed circuit board, in the upper-mid section, just right of the FG servo section.

Power Supply and Oscillator P.C. Board (CB-5002)
IdentifierSchematicRecommendedMouser Number
C51247 nF / 50 V647-QYX1H473JTP
C513100 uF / 10 V100 uF / 16 V647-UFG1E101MPM
C51410 pF / 50 V810-FG28C0G1H100DNT0
C51610 pF / 50 V810-FG28C0G1H100DNT0
C5171 uF / 50 V1 uF / 50 V FILM80-F611JG105K050L
C51822 nF / 50 V647-QYX1H223JTP
C5191.8 nF / 50 V810-FG28C0G1H182JNT0
C5201.8 nF / 50 V810-FG28C0G1H182JNT0
C521330 nF / 50 V330 nF / 50 V FILM80-R82EC3330DQ70K
C522330 nF / 50 V330 nF / 50 V FILM80-R82EC3330DQ70K
R51310 kΩ ¼ W594-SFR2500001002FA5
R514270 kΩ ¼ W594-SFR2500002703FR5
R515820 kΩ ¼ W594-MBB0207VC8203FC1
R5162.4 kΩ ¼ W594-5043ED2K400F
R5174.7 kΩ ¼ W594-5043ED4K700F
R518220 kΩ ¼ W594-5043ED220K0F
R519220 kΩ ¼ W594-5043ED220K0F
R520270 Ω ¼ W594-5043ED270R0F
R521220 kΩ ¼ W594-5043ED220K0F
R522430 kΩ ¼ W594-MBB02070C4303FCT
R523430 kΩ ¼ W594-MBB02070C4303FCT
R5241.2 kΩ ¼ W594-5043ED1K200F
R5251.2 kΩ ¼ W594-5043ED1K200F
IC504TC9142POrder from 3rd Parties
IC505BA15218NOrder from 3rd Parties
X5014.07010 MHz4.096 MHz774-MP042A-E

As always, do not hurry, take your time and do the job once. And do it well.

Article #4 | 09:04 PM Friday 11/03/2022

AKAI GX-95: Quartz Locked Tape Speed Control

This unit comes from the factory with an adjustable FG servo tape speed control. While it is pretty much bulletproof, quartz locked tape speed control is still a desirable feature. Since the system control printed circuit board in this cassette deck has provisions for the quartz locked servo tape speed control circuit, the conversion from FG servo to quartz servo is straightforward.

First, I completely removed the system control PCB from the machine. The difficulty at this step was to desolder all command and control wire ribbons. But with some care and a good vacuum pump, I removed every connection point in less than ten minutes. The FG servo control circuit is located on the left side of the unpopulated quartz servo section.

Then, I removed all FG servo related components and populated the quartz servo circuit section with the expected parts.

I have one mention, though. The solder side of these AKAI PCBs is coated with some kind of flux lacquer that emits some stinky fumes while heated. First of all, I used a combination of isopropyl alcohol and high purity acetone to clean the lacquer from the circuit board zones that I worked on. This operation is not mandatory, but I like to work in a clean way.

Here is a close range picture of the quartz servo tape speed circuit. I have used high quality components such as precision metal film resistors, quality film capacitors, C0G MLCC parts, and Texas Instruments integrated circuits. The quartz locked control IC is made by Toshiba and I bought it from eBay. I got the BA15218N high slew rate and low noise operational amplifier from the same source.

Finally, I reinstalled the system control circuit board back in its place. Those familiar with this cassette deck will quickly observe that I installed a small heatsink on the LB1649 dual bidirectional motor control integrated circuit. This part gets pretty hot during normal operation and although I'm sure it works as expected, adding a small heatsink could not possibly harm at all.

I tested the whole system and it worked perfectly. Tape speed is a bit faster than expected due to the unavailability of a 4.07010 MHz quartz crystal. The closest match I could find is a 4.096 MHz quartz crystal which gives a +10 Hz tape speed deviation.

For the record, I have found a Canadian company that can make quartz crystals to user specifications. I will try to contact them and see whether I can obtain a 4.07010 MHz part.

A particularity of the quartz servo circuit board is that upon pressing the PLAY button twice, the tape speed is halved and the PLAY triangle symbol flashes on the vacuum fluorescent display. This is an interesting behavior. I haven't tested if I can make recordings at half tape speed. Probably not, but I will still verify.

That's it for now. The conversion was a success and everything works as expected. In fact, as I write these lines, I am listening to a 90 minutes SONY UX chrome class audio cassette with some '90s disco selections. I previously recorded this tape with the Revox B215 machine. There is no audible wow & flutter at all.

For scientific documentation purposes, I measured all electrolytic capacitors that I extracted from the FG servo circuit. I used the HP 4276A LCZ Meter to collect the data in the following table.

Nippon Chemi-ConElectrolyticRadial100 uF / 10 V91.9 uF1.70 Ω
Nippon Chemi-ConElectrolyticRadial22 uF / 10 V21.67 uF6.14 Ω
Nippon Chemi-ConElectrolyticRadial22 uF / 10 V20.20 uF7.26 Ω
Nippon Chemi-ConElectrolyticRadial22 uF / 10 V19.56 uF7.46 Ω
Nippon Chemi-ConElectrolyticRadial1 uF / 50 V1.03 uF40.02 Ω
Nippon Chemi-ConElectrolyticRadial470 nF / 50 V479 nF140 Ω
Nippon Chemi-ConElectrolyticRadial220 nF / 50 V230 nF361 Ω

Immediately we observe that the capacitances are well within their specified range. However, the ESR is on the high side, especially for the small (nF range) electrolytic capacitors. As a matter of fact, I took the measurements at 120 Hz. At higher (kHz range) frequencies, the ESR drops considerably. But anyway, the high ESR is a drawback of small capacity electrolytic capacitors.

Article #5 | 08:13 PM Sunday 13/03/2022

AKAI GX-95: Impressions & Ideas

For the last two days, I've been intensively listening to various cassettes on this machine. I totaled around 20 hours of almost continuous playback. I played mostly 90 minutes BASF chrome cassettes. I have never used it for so long since I bought it more than 15 years ago.

Since I know how a good cassette deck should sound, I would dare to say that the AKAI GX-95 is definitely over average. The mechanical block is very stable and the quartz locked tape speed control is surely contributing to this feeling of quality. The cassette door cover gives a flimsy feeling, especially once the machine heats up and the dampening grease losses its grip. I like that the display can be completely turned off so that it doesn't distract me when I'm working at the PC, late at night. In addition, turning off the display contributes to preserving the electro-fluorescent elements in the VFD tube. However, I already see an issue here. Turning off the display doesn't also turn off the filaments. In the long run, I am thinking about adding a circuit with a relay that will disconnect the tube filaments completely.

I also like that I have the original remote control unit which really helps when I want to quickly pause or rewind the tape without getting out of my chair. I like the Super GX heads and I expect that I can use this deck basically indefinitely in my study to review old cassettes and listen to my various music selections.

What I don't like is the headphone amplifier, but that's pretty normal. I won't be using the integrated headphones output for more than testing purposes. My headphones are a difficult dynamic load for a simple amplifier implementation like that inside the GX-95. I will be using the RCA output to drive my main amplifier which, besides doing a good job with speakers, is also fully able to control my headphones. I also don't like that the heads are not able to extract the audio information on the tape, with the same warmness like LaserAmorphous heads, for instance. But that might be a combination of reproduction head and its amplification stages. Anyway, my ears quickly adapted after a couple of hours of listening through headphones.

Overall, I think the AKAI GX-95 is a good candidate for my primary intensive playback machine. The electronics seem to be reliable and the mechanical block is fairly easy to service. The brushed DC reel motor might pose some problems, but it appears to be of a standard type. I haven't investigated it but I think this kind of motor is still being produced. Idler tires are readily available and pinch rollers can be made to specifications. However, once the capstan bearings wear out, the mechanical block becomes irreparable. I have preventively cleaned and oiled them with Isoflex PDP 65.

For now I will continue to evaluate this deck for the next couple of months and then come back with more feedback.

your help matters

Please note that all the work presented herein this site is non-commercial. This is my hobby and I am doing this in my spare time. Through this page I freely share my knowledge with you. But if you like my work, please consider helping me buy a transistor or a capacitor for my projects.

Thank you!

Copyright © 1998- Alexandru Groza