EngineerNate's Sony TTS-4000 Build

One thing that I haven't seen mentioned is the temperature compensation effect of the two 'normal' diodes in series with the zener.
If you look at the this paper it explains the reasoning- https://www.edn.com/design/analog/4334638/Designing-a-zener-diode-regulator
When you heat or cool the zener you do it in isolation, what you should be doing is applying the heat/cooling to all three components together, only then can you see the actual effect.
Zeners below about 5.6V have a negative temp coefficient, those higher have a positive coeff due to the avalanche effect. I guess Sony picked the setup to minimise the overall effect.
You can probably buy modern equivalents for D3,4 and 5. which will reduce the effect, but probably not by much.
The engineers in those days did know their stuff. I worked with some of them.

The main reason I've considered using modern voltage regs/refs is that I figured the engineers knew their stuff and did the best with the available tech, and that aside from putting things back into spec, the greatest opportunity for long term improvement was in leveraging modern ICs that no one had access to or even conceived of in the late 1960s when this design was probably conceived.

It's interesting to me that Sony calls out a standard resistor where an NTE thermistor was actually installed. SM oversight or last minute attempt to fix drift that other compensation methods (the series diodes mentioned) failed to cure?

I honestly think the primary advantage of quartz lock or other purely frequency based control schemes is less in absolute accuracy (for to my ear, standard servo referenced tables can have perfect pitch to the human ear) but in partially decoupling the feedback loop from variances in components, whether from age or initial tolerances. As long as things don't drift so far out of spec that the circuit ceases to function, the reference frequency and corresponding output frequency from the table are nearly fixed reference points that remain unaffected by temperature drift. If the components in the servo drive drift over time or temp, the servo feedback loop can simply compensate to keep the frequency comparator happy.

Cheers,
Nathan
 
No. It isn't.

One piece of equipment gives a verifiable, industry standard measurement of wow, flutter, weighted and unweighted to JIS, DIN/CCIR. This is how the turntable was specified.

That is what Nate needs to verify performance against specification.

A polar plot achieves none of that. It takes an instantaneous reading of frequency and plots it in a 360 degree circle. By its very nature, unless you have an absolutely perfect, concentric spindle and perfectly flat record and concentric hole with respect to the groove spiral, you have a combination of compounding errors and cancelling errors in each rotation.

Many of the 'test' records are poorly pressed in that regard with little or no verification or specification of the residual W&F in the master lathe and eccentricity tolerance of the centre hole.

And all those issues affect your W&F meter as well. Also don’t forget about things like arm/cart resonance.

No doubt verifying a number is a good indicator, and no doubt some experience reading polar plots is necessary, as is acquiring plots from many ‘tables with the same test record to understand what issues are manifesting from the record, though there are tell-tales.

The punchline is I’ve seen plenty of ‘tables with reasonable W&F numbers compared to spec where actual issues can be seen on a polar plot.
 
Last edited:
SM oversight or last minute attempt to fix drift that other compensation methods (the series diodes mentioned) failed to cure?

From what I have gleaned there was a known drift issue from the get go.
As per an early review consequently lost to the ages 2250.
 
I ordered two more multimeters so I can do a better job of monitoring different circuit elements at the same time.

I've been doing a lot more reading on Vref devices to make sure I don't make the same mistake again. I will order new diodes and zeners to get it back to the reference state (though I may implement the linear reg on the 12v rail right away, I have it and that's a foolproof mod that betters the circuit) but I'm still of the belief that a properly designed vref drop in circuit is the biggest thing that can remedy any drift issues long term.

Cheers,
Nathan
 
Just an update, I played around with the old zener + series diodes, and it seems like the original zener is indeed quite weak. No load, it was sitting at nearly 8v but under load it was between 7.6 and 7.7v. The series diodes seem to be in better shape and drop ~0.6 somewhat consistently. With constant load it did vary some with temperature (cold temps seemed to have a larger effect than hot, but I just used a hair dryer and not a heat gun to be safe) but that zener voltage should be stable under load. I was running a ~12mA load across it which should be no problem.

Adjusting the load and observing the voltage, it seems the Sony IC is pulling ~15mA from Vref. A resistance of ~626ohms resulted in 8.9v across the diode cascade, which works out to roughly 14mA. Since I was seeing 8.8v in circuit it's likely the draw is slightly higher.

Cheers,
Nathan
 
Last edited:
Progress!

Work completed:

Replaced parts:
-All electrolytics. I put the two tantalums called out in the SM back to tantalum (they were called out as tants but 'lytics on the board). I also went with the 10uF SM called out caps at C5/6. Thoughts on whether those are critical or if I should try lower values like what was installed before?
-Both trimmers with 47k Bourns cermet pots.
-The vref elements. D4, D5 with EGP20A and the zener with a 1N4737
-Diode D8 for good measure. Replaced with EGP20B. I need to measure the voltage drop across the EGP20A in circuit before I trust them for the snubbers, but my plan is to replace all of the diodes.

To my knowledge, mylars and solid aluminum caps are basically immortal, so I left those alone. I've ordered Vishay Dale 50ppm 1% metal film resistors to put in as well, but I want to do things in stages to avoid causing an issue with a change and then not knowing which thing did it.

Upgrades:
-Cleaned up some of the wiring to the board and added new heat shrink to prevent shorts.
-Replaced 12v zener shunt regulator with TI TL780-12 series linear regulator.

Building the regulator circuit:
xykF5Gl.jpg

.22uF ceramic bypass on the output and .33uF on the input.

Heat shrunk and installed in board with heat sink. Thermal paste on sink as well.
MiczdTJ.jpg

UedgREr.jpg

SEBuC3a.jpg


Quick pic of the new pots + diode cascade:
PH7kGm0.jpg


After double checking things I brought it up on the DBT and I managed not to bork anything! :D The 12v rail comes up to 12.00v almost instantaneously and over the short amount of time I monitored it (I've been working on this for the last 6 hours and need to eat...) it was very stable.

I am still working on designs for the vref upgrade, and I'm not 100% happy with my zener setup. It's setting at 8.4v instead of 8.8 as the zener is at the low end of it's 5% tolerance and the new diodes drop a bit less per diode than the originals. I chose the diodes after comparing spec sheets and the EGP series seemed like the descendant of the 10D-05/10D-4 series that was in there originally but the EGP10 series is derated to 1.0A from 1.5 in the 10D series so I bumped up to the 2A capable versions. I'm not sure if that affects the forward voltage drop much.

Huge shoutout to @Pio1980 , @tnsilver , @restorer-john , @maxhifi , and @totem for their help and encouragement. :)

Cheers,
Nathan
 
Last edited:
Running it in. I'm getting some hunting here and there that doesn't seem related to the vref voltage, which is holding much steadier than before, if at a slightly lower value. Might take C5/6 down to 4.7uF to see if the lower capacitance value that was in there before was intentional.
 
Did some measuring and went ahead and replaced the remainder of the diodes. They all had black leg syndrome so it's a peace of mind thing more than anything else. I also switched in the 4.7uF caps at C5,6. They work as DC blocking caps for the input to the IC from the frequency generator. Lowering the value will raise the high pass filter a bit, hopefully reducing noise in the input. The mismatch between the table and the SM is still interesting to me. I really want to get a scope on this table so I can see what's happening at each stage.

I'm going to let it run overnight and check how it's behaving after the caps get a chance to break in a bit.

I forgot to add earlier, caps are Nichicon UKL everywhere except the snubber, which is a Panny EB and the main filter cap which is a Panny FR.

I've noticed that it "locks up" with more authority now than it did before. It almost seems to "snap" to speed. Fewer rotations to get to speed as well. I'll test all of the caps I pulled to see if one is faulty. So far they've measured okay, though I know behavior on a meter isn't always representative.
 
Checked all of the caps I removed. All measured okay except the main filter cap on the DC side, C15. It was out of tolerance on the low side. Unsure what effect that would have of the circuit considering the large amount of filtering present on the other side of the zener regulator.
 
If you want to raise the 8.4 closer to 8.8, add another series diode. Have you tried the turntable actually playing a record? Does stylus drag appreciably affect speed? I wonder if you aren't approaching the limits of what the original design can do, in terms of stability. My TD-124 drifts a bit too, it never stays locked on speed like a modern unit, there's just too many variables at play and no speed reference. It doesn't at all impact how much I enjoy using it though, I use the strobe to set the speed rather than constantly monitor it.

The crystal controlled designs which came later really tightened things up.

I'm presently restoring a Pioneer PL-L800, and it's incredible to me how much technology is in it versus the old 60s and 70s stuff I'm used to.
 
If you want to raise the 8.4 closer to 8.8, add another series diode. Have you tried the turntable actually playing a record? Does stylus drag appreciably affect speed? I wonder if you aren't approaching the limits of what the original design can do, in terms of stability. My TD-124 drifts a bit too, it never stays locked on speed like a modern unit, there's just too many variables at play and no speed reference. It doesn't at all impact how much I enjoy using it though, I use the strobe to set the speed rather than constantly monitor it.

The crystal controlled designs which came later really tightened things up.

I'm presently restoring a Pioneer PL-L800, and it's incredible to me how much technology is in it versus the old 60s and 70s stuff I'm used to.

I agree, the movement isn't all that much different from that on my CEC non-quartz servo, which sounds great to me. I haven't finished the plinth yet so no, I haven't heard a record on it. I need to fab up a temporary arm board for it. Might try to do that this week. I guess I was mainly wondering if the lower voltage might not be driving the internal transistors on the Sony IC hard enough to stabilize the speed with the correct authority. I was planning on trying just what you mentioned today. I have 3 or 4 extra diodes here from my order since the 10 piece price break made sense.

I noticed something odd in the circuit. In @mackat 's PS-2251 thread, Flavio81 noted that his table had a 2SD69 instead of the 2SC806A in the primary "throttle" transistor position. Mine has the same one! Way off the spec of the one in the SM. I'll have to read a bit of more of that thread to see what his conclusion about that was.

Opened her back up to do a diode check on all of the transistors and start the resistor replacement. I started with the control box. 1% 50 ppm 1/2w Vishay Dale CMF55's. They're so much smaller than the original carbon films, it really cleans the circuit up.
o7ZiqJb.jpg
 
Diode tested all the transistors, they're measuring okay from what I can tell without lifting legs. Some of the measurements that were supposed to have "over limit" readings did not, but likely because of other components interacting.

I added a diode to the vref chain and finished up the resistors today. Vishay Dale CMF55, 1/2w 1% 50ppm/*C metal films.

Pics:
EqJHKBD.jpg

TVHvIGv.jpg


After thoroughly cleaning both sides of the board and blowing it out with canned air to make sure I didn't have any debris to cause shorts, I brought it up on the DBT. Success! Spun right up. I adjusted the pots to get the pitch control roughly centered up for both speeds and measured the reference voltage. 8.84v, nailed it!

I'll let it run for the next twenty four hours or so to break things in and check the voltage after it's heat soaked. Then I'll check all the test point voltages and make sure we're kosher. I'm looking into getting an inexpensive scope just so I can learn more about what this thing is doing. This has been such a great project both fun wise and for learning. I can't wait to hear it play.

It almost seems a shame to leave the mylar and solid aluminum caps at this point. They're the only 6 original passives left on the board aside from the special double rectifier diodes. :D
 
Alright!

This weekend I'm picking up the wood for the plinth. I could keep fiddling with the table itself forever but I could drive myself nuts trying to get it to perfection and risk damage to the table in the process. Putting the regulator for the primary power was a no brainer but for now I think I want to leave the zener vref well enough alone until I get it up and running and can actually measure if the drift is significant.

I actually left it running overnight the other day and when I came back out the strobe was still very nearly at a standstill... so WAY less drift than with the original zener, even though I've removed the thermistor. The voltage had changed less than a tenth of a volt from initial turn on, something like 7 or 8 mV if I recall correctly. Not modern vref territory but an order of magnitude better than the 0.06v I was measuring with the original zener in place.

I'm also thinking maybe the extra diode in series is helping to improve the stability. Since each diode is probably wasting less power as heat than the zener I'm guessing they'll actually change temp a bit less, and from what I can calculate with two diodes in series the tempco of the zener should be theoretically balanced, but since I don't have the ability to directly couple their temperatures the extra compensation from the third diode may be helpful. This is my plausible explanation for why removing the thermistor hasn't caused additional drift.

As a last update on this table, I purchased a Keystrobe Pulser PCB and the SMD LED circuitboard from John's Garrard 401 kit and I have installed that in place of the stock strobe. My goal with this restoration/upgrade is to do as little harm/permanent modification as possible. I want this thing to be able to go back (functionally) 100% to stock if I or someday it's next owner wants that at some point in the future. so my process was as follows:

1. Remove strobe assembly and carefully remove neon lamp without damaging it. It was glued in, so I got lucky here.
2. Remove neon wiring from harness. This was simple enough, just slip the two contacts out of the big molex. This way I don't have to cut the stock wiring. The neon + wiring will go into storage.
3. Wire up LED PCB and install:
dhdBHCV.jpg


LTIwpyj.jpg


rfRCYfi.jpg


I did RC for years so I have a stockpile of deans connectors. If they can hold while I'm doing rolls and loops they'll hold inside a stationary table. Super low contact resistance as well. Wire is teflon insulated silver tinned copper I got in bulk from eBay. Super awesome stuff to work with.

4. Wire up and Install PCB. This is a prototype made from old hotel door cards. I want to order some GF10 to replicate the mount and put screws through it instead of just a dab of hot snot.
QUmgQq4.jpg


I tied into the AC output from the transformer as to not add any load/noise into the DC side of the control circuitry. The Keystrobe Pulser has a built in rectifier that can do up to 20VAC so this was perfect. It was also much easier to tap into the AC side of things on the circuit board, I just tapped right into where the AC enters the board:
mhZuCsk.jpg


5. Test it! It works!
iSvdBVV.jpg



Incredibly bright, incredibly sharp strobe picture. Huge props to John at Keystrobe for a great product and for his great tech support via email.

I've sealed her up and I'm not doing any more mechanical or electrical work on the table until I can listen to it and evaluate if any of the things I've noticed while refurbing it are audible. Can't wait to get my hands dirty with the plinth build. :)

Cheers,
Nathan
 
BTW, for the curious, this is what the FG output looks like on a scope. I just spun it by hand for this so don't read anything into the magnitude/frequency. Nice clean sine output.

TAHOMKs.png


I bought my own USB scope and at some point I'll get shots of all the test points, but I'm getting a nice stable speed out of it so I'm inclined to leave well enough alone for now.
 
Finally sorted out an azimuth issue with my AT1005ii and got some plays in with the test armboard.

sOoQg4n.jpg


Switched over to my AT95HE stylus after a test side. Sounds fantastic. Speed is stable, presentation is smoooooth and detailed. Zero rumble I can detect.

Cannot wait to get my plinth parts back from the laser cutter. They had a death in the family that may have slowed things a bit. My wife is out of town next weekend so I'm hoping to get it this week so I can distract myself with the build. :D

Cheers from a happy listener. :music:
 
Almost forgot. I got some 1/16" G-10 fiberglass to make a permanent mount for the Keystrobe. Just need to get it cut to shape and install the board standoffs on it.
 
The laser cutters sent a present home with me today. :jump::banana::banana::banana:

Packed carefully into the car...
HT1HIzs.jpg


Stacked in order:
yM6STCU.jpg


And armboard fitment:
HHxG2D4.jpg


So incredibly stoked. This thing is MASSIVE. I knew it was going to be big to handle 2 x 12" arms, but HOT DAMN. I'm glad I've got my stereo rack in progress as otherwise there would literally be nowhere to set this thing... it's far too large for my current end table setup.

The laser cutter did an amazing job and they were super easy to work with. For anyone in Eastern Tennessee/Western VA, Virginia Laser in Abingdon is top notch and very reasonable price wise for low volume work.

I wanted to drill all of the armboards to make their outer edge line up with the table, so I had them punch those holes as a simple pilot. I'll clamp them up to the first sheet underneath with the holes predrilled and punch them through.

I had two of the armboards cut to match my Mayware and AT1005ii arms and 4x blanks. Six armboards should be more than enough for now, especially if I drill one to take a Jelco type plate mount and another to take the SME mount.

Unfortunately (for my build :D) I'm flying to SoCal tomorrow for my brother in law's graduation, so I'll not get started on this until next week. I also want to try to find a slightly larger hardwood dowel to use, the "1/4"" that I got from Michaels seems undersized significantly and is quite loose in the locating holes for glueup.

Cheers,
Nathan
 
Back
Top Bottom