F2624 Resurrection: A modern remake of a vintage classic

Status
Not open for further replies.

tnsilver

Stereo Puppy
A year ago I encountered a unique DIY project. This talented electronics major from the University of Bar-Ilan in Tel-Aviv had built his very own CNC machine. I was amazed. I told him that if I fail to google up somebody who DIY'ed a ballistic missile in their own back yard - his is the coolest project I've ever seen.

I was struggling, at that time, with a 9090DB unit I picked up in a tag sale. I was stuck with a faulty F2624 driver amp board. It's rails lifted, it had too many DIY jumpers, missing solder rings, bad rails under the Molex assemblies, shot trimmers and it was giving me hell.

I've asked Anatoli, the owner of the CNC machine, to carve me up a new F2624 board and that made him laugh. As it turned out, he could actually do it, but he said there's a much better way. Being an Electronics major and a university technical assistant, Anatoli had both the knowledge to re-design the board plus the access to the university labs and testing gear. He also had the right designing software. Almost a year later we ended up with this:

The new F2624 Driver amp board

F2624_FRONT_1.JPG


First thing Anatoli asked for was the schematics. You all know how ridiculous and erroneous those are. It seems the 9090DB service manual was made to fool technicians. It's a circus of a document. As a result, Anatoli had to reverse engineer the F2624 board I sent him and over time, he came up with his own schematics.

We started to look for modern replacement parts. The board's 70's components all have modern replacements. Those are light years advanced than their 70's counterparts but some components, like the VD1212 red dotted rectifiers and the Temperature compensating STV-H3 diodes are no longer available. We also had to locate some suitable heat sinks and Molex connectors. We then had some boards printed. It took the better part of this year (2013), and finally... It was done. I got the first working board last week.

Fresh F2624 PCB's

F2624_AND_PCBs_1.JPG


I put my tag sale 9090DB on the workbench, plugged it to a DBT and installed the new board. Anatoli had tested the newly fabricated board in his lab to his very own satisfaction but all that time, he did not have access to a working 9090DB unit. This board had yet to prove it's worthiness.

I was terrified of the PITA bias procedure, This very gig caused me to blow my outputs more than once in the past, by overshooting the trimmers setup. I was very careful this time. We've discussed this procedure in length and paid a lot of attention to this issue. To my surprise, it was rather easy this time. I waited the 4 specified minutes for the unit to heat up and started adjusting the bias towards the 30mA mark. Very stable, very consistent, very easy - the bias current went up to ~30mA and started to drop to the 29.1mA after exactly 4 minutes, where it stayed fixed. I was enthusiastic.

My $25 tag sale purchased unit had some mods done on it , including the replacement of the OEM filter caps to a pair of Nichicon KG 15,000uF/80V 100mm long Super Through's and I could not be any more happier with the result. It was still on DBT but I've never heard a DB sing like this before. Crystal clear, punchy lows with the land mark richness of the famous 9090DB warmth plus endless power reserves. I was freaking high!

Initial F2624 Driver Amp Board Testing

9090DB_NEW_F2624_TEST_3.JPG


This week the unit was put on mains. I just can't get enough of it. Both phono inputs work flawlessly, the AUX entry is spec-less and even the overhauled Dolbey board plays it's test tune with both meters reacting evenly and when no source is playing, the unit is dead silent (even at near max volume).

Another shot of the new F2624 board

F2624_UNDER_1.JPG


So, why am I so proud of this board? Well, first of all, it's one hell of a vintage resurrection project. Second, Anatoli and me defined some objectives prior to getting started, such as the complete ban of low quality cheap Chinese components. It's all high end electronics here now. Nichicon EC's and MUSE bi-polars, Murata MLCC's, KOA Speer resistors, ON Semi-conductors, Zetex transistors and other quality components. Then, we wanted to correct all that is wrong with the OEM board.

You all know how the rails lift when you over heat them by accident when leaving the tip of the soldering iron on them for too long, and how those flimsy soldering pads tear and detach from the board. There is no one here who ever serviced one of those boards and could not stop second guessing the lousy silk screen. You just can't trust it.

With the OEM silk screen, diodes directionality is marked wrong, some caps polarity is either missing or wrong and those transistors ECB markings are so missing on the back of the board, it's crying out loud. Most of us, who serviced that board, probably had to resort to some kind of an improvisation to adjust newly ordered replacement caps lead spacing to the holes on the board. I haven't mentioned yet the need to serial solder two 1N4148 diodes in mid air to replace the VD1212 diodes or the annoyance in having to sort out the Zetex ZTX694B pin out to match the 2SC1400 differential pair transistors. In short - the OEM board rebuild is a pain! All of this hardship - is now eliminated! However, our most important goal, our prime directive, was to not alter the legendary 9090DB vintage sound!

F2624 Front (Component Side)

F2624_PCB_FRONT_1.JPG


List of features in the new F2624 driver amp board

  1. Double layered PCB. All wiring is done on board, no need for air wiring components

  2. Dual footprint for ceramic caps will fit both 2.5mm and 5mm lead spacing

  3. Dual footprint for EC's will fit 2 sizes as well (8 - 10mm diameter with 3.5 - 5mm LS and 16-18mm diameter with 5-7.5mm lead space)

  4. Trimmer footprint will fit the original trimmers (5mm between wiper pin and the two other pins) and will also fit modern substitutes with 2.5mm spacing between wiper pin and the other two pins.

  5. All parts are placed on top of the board as well as R901/902. Placement under the board is obviously still possible if anyone insists on doing so

  6. D05/06 (VD1212) footprint now allows the use of 2 1N4148 plain Jane rectifier diodes in series directly on the board. No need for air soldering.

  7. Fits the original connectors, no need for any changes to other parts - just plug directly instead of the original board.

  8. All parts other than the Zetex differential pair transistors are pin compatible with the stock parts. The heat sinks are of slightly larger size than the OEM counterparts.

  9. The thermal diodes (STV-H3 replacement devices) had been selected after testing the temperature coefficient in a temperature chamber at the Bar-Ilan university lab and after selecting the best matching modern equivalent components for the originals.

  10. Detailed silk screen on both sides of the board. The top side contains part numbers, EBC marking for transistors, polarity marking for capacitor and diode directionality. The back side contains polarity marking for capacitors, pin numbers on all headers and EBC markings for BJT's.

  11. All parts (other than the thermal diodes) are readily available for purchase with concurrent mass suppliers (DK, Mouser etc...).

  12. Use of BZX55 low noise zener diode as replacement for the OEM Zeners

  13. Audio polar capacitors replaced with audiophile level Nichicon MUSE bi-polar electrolytic capacitors and it's possible to use film caps too - if you choose to.

F2624 back (Conductor Side)

F2624_PCB_BACK_1.JPG



Anatoli and me are planning on placing the new F2624 board back in production. We can manufacture them at an impressing rate. We did not, however, plan any marketing strategies yet so please don't ask for quotes and discounts yet (although I'm gonna push for an 'Exclusively Sansui' membership discount. :yes:).

Any comments are welcome. Feel free to ask anything (no answers are guarantied) and most importantly... Get jealous guys. The more the better. We've earned it :wave:

In the mean time... here are some more pics...

F2624 Both Sides

F2624_PCB_BOTH_SIDES_3.JPG


F2624 Old Vs. New

F2624_OLD_VS_NEW_1.JPG


Cheers

Tom
 
Last edited:
[*]The thermal diodes (STV-H3 replacement devices) had been selected after testing the temperature coefficient in a temperature chamber at the Bar-Ilan university lab and after selecting the best matching modern equivalent components for the originals.
...and you ended up using **what** part in its place??
 
...and you ended up using **what** part in its place??
:thmbsp: Now that's the kind of proffesional attention I was expecting from the prominent AK techs... I'm glad you brought it up.

It's all based on your wonderful work right here. The STV-H3 replacement device in our new F2624 was constructed in a very similar way to the device you constructed using a series of 3 rectifier diodes layered in a solid thermal conductor compound (actually the same commercially available type you've referred to, and using the very same plain Jane rectifier diodes you've tested).

The only real difference between the device we've formulated and your prototype is the termination style. I think the way we've designed it, assures a little more surface contact with the transistor heat sink.

Anatoli, my colleague, wanted to be sure about the device so he conducted his own temp-co tests in a university lab temperature chamber. His results, very much corresponded to your own and we finally chose the same type of diodes (which are also very cost effective when bulk ordered).

I hope you're not mad about it - we've been extracting every shred of info you've contributed here in order to make this board materialize and we are greatly in your debt - as well as for many other prominent AK members. We appreciate it very much!

Here are some related pics...

The cast

n_8_w_X95z1lnsUWLJPD_0jZHihQyHsS5sxVm45NaRM=w647-h483-no


OEM Vs. Remake

STV-H3_MAKING.JPG


Cheers

Tom
 
Last edited:
DAMN GLENN! I wanted to ask that question. :D

I like how he mounted the replacements on the heatsinks of the inboard TO-220's.

Larry
 
This is some truly superb gifted r&d work - Period!

I am sure there will a good few 9090DB owners who will be contacting both you and your colleague in the near future to get hold of your F2624 boards! :D - (either pre-built or kit form)

Like the STV-H3 replacement too! - [I can see a queue forming for just these alone! - and maybe you could do the same to replace an MV103 ?]

Keep up the superb work :tresbon:
 
Last edited:
Yeah I was going to ask about the STV diode as well.....Great job on that by the way, it was one of the first things I noticed on the board.
I like the way you molded that, very impressive.

BUT, seriously dedicated and thorough effort from you guys, making full use of the ability to make a multi layered board, which I am sure the original Sansui tech's would be very impressed with!!
This is the part that most impresses me too......

Just brilliant guys and well done!!........:thmbsp:

I would go for three different options, 1) sell the board only, 2) sell the board + components as a kit-set, 3) Sell the complete board loaded and tested.

GREAT WORK......:thmbsp::thmbsp:
 
Thank you people. Thanks a lot for the kind words! These are the kind of responses we were opting for. That board was designed with many AK'ers F2624 issues in mind and it aims at solving many of the OEM problems. In short, it was designed for you - if you're reading this.

Kevzep, your kind words are very reassuring and coming from down under it's a big complement for us. We already got references from the continental US and the UK and yours makes it truly globally exciting. Thank you. We appreciate it. Your marketing suggestions had occurred to us too (exactly as you've mentioned them) and we'll formulate pricing and offering pretty soon. We are probably much better at vintage gear than selling boards and we need to take our time. :D
 
Last edited:
Congrats , this ia a masterpiece. The whole thing is a masterpiece.
From conception to execution.
Have no words.. Excellent.
 
Beautifully executed and what an honorable concept it is too build upon the great legacy of these masterpieces. I truly wish you the best of luck!
 
Nicely done. I'd have never guessed that there were three 'UF' diodes in that casting.

Yes, I was also surprised at how well the 'UF' diodes follow the original STV diodes when warmed up. Anyway, that's what I posted the information for. Glad to have helped out.
 
Very good job!!
Cheers to Anatoli, your colleague as well.
BTW, it's STV-3H :D
What, I just looked at the sch of the PA(F-2624)
Vbe bias multiplier is shown using
Tr9,10(2sc984b,c)
and your pic shows them mounted on one driver bjt tab(HS) = strange.
Usually you put thermal sensor on the big HS, so that bias tracks them not the drivers?. Some put the drivers on the big HS as well, so that they all track to the same relative temp.
So you guys decided to change the bias arrangement to a STV-3H, okay, but why?
Did you change the o/p's bjt's, and to what MJ2119x? Nice to see an old TO-3 design.
D05,6 is just a double diode, nothing special. 1N4148 x 2 would work fine as a bias generator for a CC source. or an LED.
Interested in doing more work on this unit?
I looked at the phono, uses a BA312. An old Rhom part, have to look it up, okay a bjt array. Sound any good, noisy?
Looking to upgrade any thing else?
 
We didn't change the design. It was our main goal. The STV-3H choice was made long ago by Sansui. Perhaps you're not aware of the multiple versions of the F2624 board. An early OEM version of the F2624 did use 2SC984 transistors for that purpose, the later used STV-3H diodes which are unobtainable now. The service manual was never updated and it's a miserable source for an historic account of the board's mutations over time.

The two 1n4148 diodes in series are a very common replacement for the OEM VD1212 diodes that are no longer obtainable either. The arrangement on the board merely saves the annoyance of soldering the little Fairchild's 1N4148 together and sticking them in the tiny lead space holes gap on the OEM board. It's a common issue for 9090DB owners.

We did replace the old Zeners with the less noisy BZX55. The unit is dead silent. Totally silent even at near full volume with no source playing. Off course, it was also properly serviced other than the new driver amp.

We're happy with this board. We're not planning any changes and we never really performed any. My name is Tom, my colleague is Anatoli and all of the board's semi-conductors had been replaced with modern equivalents that are very common on this forum.

Cheers

Tom
 
Last edited:
Tom and Anatoli, this is fantastic work! I have a QRX-9001 waiting for me to get to it that contains two of these boards. When you're ready to sell kits or complete boards, let me know. Tom, if you're in Boston I'm right up the road in Arlington. Thanks for this huge contribution to the "Keep the old 'Suis singing" effort. --Dan
 
In my humble opinion that is exceptionally well done !

Congratulations Tom & Anatoli - very impressive work.

I look forward to further news and developments.
 
In my humble opinion that is exceptionally well done !
Thanks John, it's a pleasure to hear that especially coming from you. Thank all of you, including my Arlington neighbor. First batch of Mouser components on it's way, we'll have some boards available soon...

I think Anatoli is on a trip to eastern Europe but he's got an AK account and he sure reads this thread. So when he can, he'll post something about it too. In the mean time, here is a pic I ripped off his post, showing his early F2624 tests:

f2624_ht_4.jpg


Cheers

Tom
 
Last edited:
we'll have some boards available soon...

From the AK rules:

Advertising:

No advertising of any kind (even give-aways) is permitted on AK except under these rules, which are strictly applied:
  • You cannot advertise or promote any business without pre-approval as an AK Vendor.
  • All ads or offers to sell, buy or trade anything must be in Barter Town.
  • Non-Subscribers cannot make offers to buy, sell or trade anywhere on AK.

Please contact an AK admin regarding our reasonable rates.
 
Very nice work. I'm sure people will be lining up for these boards. Used ones are going for $100 on eBay!

- Pete
 
Very nice work.
WoW Pete... Thanks a lot! Coming from you - it almost makes us blush :thmbsp: Your input on this forum was an inspiration in this project. Your series of posts about the Dolby board overhaul from '07 are still remarkable and none of this would've happened without your contributions to this amazing little community of Sui' fans.

Anatoli and myself would also like to express our gratitude to Dave (avionic), Jim (CDFixer), John (Hyperion), and off course (as we did already but never enough) to Glenn (EchoWars) and many others who deserve angel's wings for their help in spreading the Karma. Keep it up!

Tom
 
Status
Not open for further replies.
Back
Top Bottom