Sansui Turntable - Particularly XR-Q11
mootej
Active Member
This is, it seems, kind of a redundant sensor scheme to sensor A. The unusual thing I already see, the signal is pulled off at op-amp Stage 2 which is still kind of noisy. The transistor stage is, I guess, an input buffer. I had to go refresh my transistor understanding. I always liked vacuum tubes, it always seemed kind of straightforward. Transistors, while more-or-less the same, somehow always escaped me a little bit.
The transistor is in a common emitter configuration. The load line is constructed by locating the supply voltage on the x-axis (5V) and the collector current (equal to supply voltage divided by the load resistor (5V)/(3300ohm) = 1.5mA
Weird place to locate the load line. I’m, pretty sure this will be an on-off transistor configuration rather than a linear amplifier. Common emitter will also invert the signal. So the input to the transistor somes from Stage 2 which is mostly a noisy 0 to 3V with spikes at the record lead-in, gaps and lead-out that are up to 12 to 14V. See below for what the output from Signal E Stage 2 looks like.
Note that as a voltage spike occurs at the input to the transistor, the transistor base starts to conduct. This causes the transistor collector to emitter to conduct (resistance collector to emitter goes down). The collector, also known as the output, now approaches the voltage more toward the ground side of the transistor rather than the supply side. So yes, signal inversion, high input signal equals low output signal. When the signal coming in is low or near zero, the signal out will be high, near the max of the supply to the transistor. In other words, +5V. And our spikes that we want to recognize? Spikes down from +5V to 0V. This signal is then fed in to the multistable multivibrator IC9a, pin5.
Now, here is an interesting twist. Ain (pin4) is grounded (low). The signal goes in to Bin (pin5). The IC is looking for a high to low transistion at pin B in order to output a pulse. So yes, it is correctly looking for an inverted signal as presented by the transistor. Output comes out of IC9a pin6, the regular, not the inverting output. So we should get a pulse out from low (0V) to high (+5V) for 0.45sec. The pulse is sent to the MPU, pin3, input PC1).
Why do I propose a summer intern? First, picking off the noisy signal at Stage 2 seems questionable. Maybe Stage 2 output is not supposed to be so noisy? Well, I said before, that is one to watch. Maybe the phototransistors have aged and are just noisy.
This transistor input design line is also suspicious and looks really sensitive. It seems to me it will either react to the smallest signal (noise) and stay in the conducting state all the time or just the opposite, it will slam against the other extreme and stay non-conducting all the time. And lastly, all the inversion. Well, it’s available on the chips, I guess. Just confusing and, with logic tables, it can get a little dicey.
The interesting part? Vinyl record size identification seem to be working OK. The system seems to be finding the record start relatively well. I think there’s a reason for this that you’ll see later when you see the actual signal traces. The E3 sensor signal looks like it is slamming to the high side (+5V) most of the time. In this state, it does not overide Sensor A (located underneath the tonearm), and in fact, defers to Sensor A to initiate the tonearm drop onto the record start. In essence, it is not really used.
Here's what things are really doing:
So, E3, mostly at +5V, mostly doing nothing to inhibit Signal A from finding the record start.
The transistor is in a common emitter configuration. The load line is constructed by locating the supply voltage on the x-axis (5V) and the collector current (equal to supply voltage divided by the load resistor (5V)/(3300ohm) = 1.5mA
Weird place to locate the load line. I’m, pretty sure this will be an on-off transistor configuration rather than a linear amplifier. Common emitter will also invert the signal. So the input to the transistor somes from Stage 2 which is mostly a noisy 0 to 3V with spikes at the record lead-in, gaps and lead-out that are up to 12 to 14V. See below for what the output from Signal E Stage 2 looks like.
Note that as a voltage spike occurs at the input to the transistor, the transistor base starts to conduct. This causes the transistor collector to emitter to conduct (resistance collector to emitter goes down). The collector, also known as the output, now approaches the voltage more toward the ground side of the transistor rather than the supply side. So yes, signal inversion, high input signal equals low output signal. When the signal coming in is low or near zero, the signal out will be high, near the max of the supply to the transistor. In other words, +5V. And our spikes that we want to recognize? Spikes down from +5V to 0V. This signal is then fed in to the multistable multivibrator IC9a, pin5.
Now, here is an interesting twist. Ain (pin4) is grounded (low). The signal goes in to Bin (pin5). The IC is looking for a high to low transistion at pin B in order to output a pulse. So yes, it is correctly looking for an inverted signal as presented by the transistor. Output comes out of IC9a pin6, the regular, not the inverting output. So we should get a pulse out from low (0V) to high (+5V) for 0.45sec. The pulse is sent to the MPU, pin3, input PC1).
Why do I propose a summer intern? First, picking off the noisy signal at Stage 2 seems questionable. Maybe Stage 2 output is not supposed to be so noisy? Well, I said before, that is one to watch. Maybe the phototransistors have aged and are just noisy.
This transistor input design line is also suspicious and looks really sensitive. It seems to me it will either react to the smallest signal (noise) and stay in the conducting state all the time or just the opposite, it will slam against the other extreme and stay non-conducting all the time. And lastly, all the inversion. Well, it’s available on the chips, I guess. Just confusing and, with logic tables, it can get a little dicey.
The interesting part? Vinyl record size identification seem to be working OK. The system seems to be finding the record start relatively well. I think there’s a reason for this that you’ll see later when you see the actual signal traces. The E3 sensor signal looks like it is slamming to the high side (+5V) most of the time. In this state, it does not overide Sensor A (located underneath the tonearm), and in fact, defers to Sensor A to initiate the tonearm drop onto the record start. In essence, it is not really used.
Here's what things are really doing:
So, E3, mostly at +5V, mostly doing nothing to inhibit Signal A from finding the record start.
Last edited:
mootej
Active Member
F1 is exactly like E2 and needs no real further explanation. F2 is the same as F1 and E2 except there is a special input to pin3, the inhibit switch. It come as an output from the MPU through a buffer transistor. For right now, let's just see what the traces around F1 and F2 look like. I guess I'll really need to get and mount a cartridge soon since both the F signals are really meant to sense things with the tonearm down. I guess I can tackle the inhibit and muting circuits a little later when track select is working.
So we can probably make a few measurements, then hang things up for a little bit while I get a cartridge, the cartridge adapter and the replacement trim pot. Maybe I'll also contact the guy, Kevtris, about how he does a "mask-rom" test mode dump...
I also ordered some replacement E and F phototransistors and LED's. I may decide to just replace all of those components to see if the "noise" can be reduced. It may be a little bit to get all the parts in hand...
So we can probably make a few measurements, then hang things up for a little bit while I get a cartridge, the cartridge adapter and the replacement trim pot. Maybe I'll also contact the guy, Kevtris, about how he does a "mask-rom" test mode dump...
I also ordered some replacement E and F phototransistors and LED's. I may decide to just replace all of those components to see if the "noise" can be reduced. It may be a little bit to get all the parts in hand...
Excellent write up all around. And a level of detail qnd thoroughness we seldom get treated to here. I do want to ask if you are purchasing any of these parts in quantity beyond what you need, as I may be contacting you to work out a deal for some spares from your alottment (if avaiable), as well as that XR-Q9 now that the holiday craziness if behind us.
mootej
Active Member
Of course Sansuiman. I tried to order multiples where possible because shipping will absolutely kill you! And I'm certainly ready to share if it means more rescued top-end Sansui turntables. Funny, the parts are coming from all over the world...
mootej
Active Member
Well, I thought the F1 was going to be boring...
Here's the input, all good for the input (orange):
Here's a zoom-in:
Input: zero, then spike to 4.7V for the lead-in, 5-tracks/4-gaps, lead-out, then repeat in the reverse direction. But the output (blue), nothing.
My scope is a little funny, not very sophisticated, not very high resolution. So I reset from 500ms/div continuous run to 5ms/div, triggered on the input signal for a single scan:
There it is! Input (orange) looks good and output (blue) is a spike-pulse to 5V. But wait...zoom in:
The pulse is only like 80usec, well, somewhere in the 40 to 120usec, with the kind-of crappy resolution I have. Still, way short. Based on the datasheet and the resistor and capacitor, it should be like 100msec.
Here's the input, all good for the input (orange):
Here's a zoom-in:
Input: zero, then spike to 4.7V for the lead-in, 5-tracks/4-gaps, lead-out, then repeat in the reverse direction. But the output (blue), nothing.
My scope is a little funny, not very sophisticated, not very high resolution. So I reset from 500ms/div continuous run to 5ms/div, triggered on the input signal for a single scan:
There it is! Input (orange) looks good and output (blue) is a spike-pulse to 5V. But wait...zoom in:
The pulse is only like 80usec, well, somewhere in the 40 to 120usec, with the kind-of crappy resolution I have. Still, way short. Based on the datasheet and the resistor and capacitor, it should be like 100msec.
mootej
Active Member
Way too short of a pulse. What might cause this?
- bad IC chip? Probably not, it seems to be outputting a pulse.... But maybe...
- bad resistor - no, I measured them
- bad capacitor? Those pesky tantalums...
I may be re-capping sooner than expected. Bad immediate news, my capacitor tester does not seem to be working, at least with the caps soldered in. More bad news, the guy could not find the part from the FR-D55 that had the trim pot I needed. I did establish that the bad trim pot had a leg torn out which is why the resistance is above 30M-ohm. It's hanging by a wire...
Guess I'll test F2, see what we find there, then go look for some test parts in the junk pile to see if I can get things to improve, then hunt down some actual replacement parts.
And @Sansuiman, I'm hoping all this inspires you to drag out and fix all those great spinners you've got in storage, now that we know where all the soft spots are. I still want to strike up a deal for one of your lost souls so I can get some parts.
- bad IC chip? Probably not, it seems to be outputting a pulse.... But maybe...
- bad resistor - no, I measured them
- bad capacitor? Those pesky tantalums...
I may be re-capping sooner than expected. Bad immediate news, my capacitor tester does not seem to be working, at least with the caps soldered in. More bad news, the guy could not find the part from the FR-D55 that had the trim pot I needed. I did establish that the bad trim pot had a leg torn out which is why the resistance is above 30M-ohm. It's hanging by a wire...
Guess I'll test F2, see what we find there, then go look for some test parts in the junk pile to see if I can get things to improve, then hunt down some actual replacement parts.
And @Sansuiman, I'm hoping all this inspires you to drag out and fix all those great spinners you've got in storage, now that we know where all the soft spots are. I still want to strike up a deal for one of your lost souls so I can get some parts.
I want very much to get them up and running. They've been waitng a very long time; far too long. Of course life (and other projects) keep getting in the way. Right now there are three items on my test bench in progress (none going smoothly) only one of which is Hifi related. The other two are higher priority anyway, as they are equipment that I use for work. I have begun sending audio gear out to a trusted local tech jist so I can get some of the backlog knocked out. Way too much gear here sitting and waiting for it's turn. Just as an example, I have a Technics SL-P1200 desk console cd player that has a strange audio output issue dogging me, and am awaiting an original copy of the FSM to arrive from Germany to hopefully get that resolved.
I guess I need to resolve to make a Q9 or Q11 the next on the bench. Maybe if I get one going, I'll find it is not so bad a project and can get 2 or 3 done. But, then I have to find a place to use them, and presently my XP-99 and SR-929 are serving me quite well. Maybe I need to thin the herd. A working turntable of this calibre should raise a few bucks I suppose.
On another note, I'm scrounging up proper packing for secure transport of an Q9, so you'll be hearing from me shortly about that, once I have it set to ship.
Finally, I wanted to thank you for spearheading the full run down on repairing these tables, and documenting it here in such phenomenal detail. Realistically, I know I'd never have gotten as far into it as you managed to. I'm decent at troubleshooting, but not so much at documenting a lengthy process as you've done now. With this thread, you have provided everything neccessary for anyone with competent tech skills to troubleshoot and repair one of these units.
That is both an excellent accomplishment, and major contribution to the cause of preserving gear of this sort that would otherwise languish (or get chucked in the dumpster) from it's complexity, putting off any who wanted to attempt repairing them. Thr XR-Q9 and 11 have now been "de-mystified" sufficiently to make successful repairs much more likely. I appreciate and respect the time and effort spent on pursuing this to it's conclusion.
I guess I need to resolve to make a Q9 or Q11 the next on the bench. Maybe if I get one going, I'll find it is not so bad a project and can get 2 or 3 done. But, then I have to find a place to use them, and presently my XP-99 and SR-929 are serving me quite well. Maybe I need to thin the herd. A working turntable of this calibre should raise a few bucks I suppose.
On another note, I'm scrounging up proper packing for secure transport of an Q9, so you'll be hearing from me shortly about that, once I have it set to ship.
Finally, I wanted to thank you for spearheading the full run down on repairing these tables, and documenting it here in such phenomenal detail. Realistically, I know I'd never have gotten as far into it as you managed to. I'm decent at troubleshooting, but not so much at documenting a lengthy process as you've done now. With this thread, you have provided everything neccessary for anyone with competent tech skills to troubleshoot and repair one of these units.
That is both an excellent accomplishment, and major contribution to the cause of preserving gear of this sort that would otherwise languish (or get chucked in the dumpster) from it's complexity, putting off any who wanted to attempt repairing them. Thr XR-Q9 and 11 have now been "de-mystified" sufficiently to make successful repairs much more likely. I appreciate and respect the time and effort spent on pursuing this to it's conclusion.
mootej
Active Member
This has gotten way more interesting! My meter was not able to measure the capacitance in circuit for whatever reason. Gotta admit, I never really trust it anyway. No matter, I got a way better idea. OK, not for the faint of heart, Bode plots. So if I hook up the capacitor with a resistor in series as a low-pass filter and put a frequency on it, then measure the output, I can look at the amplitude output compared to input (and the phase angle). I don't want to get into it with imaginary math and all that, so let's just say we can measure the frequency response of the capacitor and see if it is working or not. I used a 180-ohm resistor. The C29 capacitor I tested is supposed to be 0.68uF. As a reference, I also tested a known good 1uF capacitor. Here are the plots, first amplitude, then phase angle:
Last edited:
mootej
Active Member
Well, I can't explain the exact failure mode of the tantalum capacitor but you can see it is nowhere near what a normal capacitor should be. So the pulse circuit is not working which means the computer can't see what is happening and the whole thing doesn't work. And there are a whole lot of these tantalum caps! And they're probably all bad!
HUGE FIND!!!
So if you have a Sansui turntable, check the tantalum caps first!
Wait, maybe I jumped for joy too soon. I forgot, the IC is still connected.
HUGE FIND!!!
So if you have a Sansui turntable, check the tantalum caps first!
Wait, maybe I jumped for joy too soon. I forgot, the IC is still connected.
Last edited:
mootej
Active Member
Here is the circuits I set up to test:
Vin came from the signal generator part of my PCSU200 (from Velleman) $50 oscilloscope and signal generator. Vout goes in to Channel 1 and provides the amplitude plot. Channel 2 is hooked up to Vin and provides the phase plot. I thought I would have to do this all manually but I discovered, the PCSU200 will do a Bode plot automatically! This thing is awesome for $50 used. One of the best purchases I've made.
I did forget, though, the IC pin 1 and 2 are still connected across Vout. Not sure how this will affect the response.
I hooked up a couple of alligator clips to the capacitor on the board, ran them out to the oscilloscope, signal generator and resistor (a 180-ohm). I figured I would get a drop-off frequency (-3dB) of around 1/(2pi(RC)) = 884Hz for the reference 1uF cap and like 1300Hz for the on-board cap. The 1uF reference plot was perfect, just as expected. Voltage right up to at 100% before 800Hz, then drop off to <15% at higher frequencies, phase angle from zero below 800Hz going up to -90 degrees at higher frequencies.
I guess I'm suspect of the cap but I guess I really need to pull it out before I condemn it.
Vin came from the signal generator part of my PCSU200 (from Velleman) $50 oscilloscope and signal generator. Vout goes in to Channel 1 and provides the amplitude plot. Channel 2 is hooked up to Vin and provides the phase plot. I thought I would have to do this all manually but I discovered, the PCSU200 will do a Bode plot automatically! This thing is awesome for $50 used. One of the best purchases I've made.
I did forget, though, the IC pin 1 and 2 are still connected across Vout. Not sure how this will affect the response.
I hooked up a couple of alligator clips to the capacitor on the board, ran them out to the oscilloscope, signal generator and resistor (a 180-ohm). I figured I would get a drop-off frequency (-3dB) of around 1/(2pi(RC)) = 884Hz for the reference 1uF cap and like 1300Hz for the on-board cap. The 1uF reference plot was perfect, just as expected. Voltage right up to at 100% before 800Hz, then drop off to <15% at higher frequencies, phase angle from zero below 800Hz going up to -90 degrees at higher frequencies.
I guess I'm suspect of the cap but I guess I really need to pull it out before I condemn it.
mootej
Active Member
I have a question for y'all. The datasheet diagram above has a diode across the Rx. What is that there for?
I tested a bunch of the tant caps and they mostly look bad-ish. I'll post some examples soon. I couldn't exactly figure out how the capacitor may have failed so I downloaded LTSpice (scary! but Free!) and fumbled through making a schematic using a resistor and a non-ideal capacitor, a voltage source and ran a frequency response simulation...took a while to figure out how to do this. Then I right-clicked on the cap and changed the non-ideal values. Well, gimme a break, we were just watching Vera reruns. The upshot, when a capacitor ESR (equivalent series resistance) increases, ALOT (like maybe to 100 or 1000 ohms!) then the frequency plots start to look a lot like the ones we get on these capacitors.
I look up the failure mode of these tant's for the case when they show high ESR. It seems it can be caused by peak currents on startup or AC ripple. So then I realize there is a diode on the datasheet and not one in the turntable circuits. Then I remember reading it is really bad to reverse bias these ELECTROLYTIC tant caps. My guess, this is some kind of design problem with something happening at either power-up or power-down and a diode would have prevented this. It's a theory but all these caps going soft? Addendum: perhaps design problem is a little harsh...a design problem to make it last 40+ years.
Looks like a recap is coming. I may try something temporary to assure myself this will fix things before I take the main board out. Looks like it will be a major PITA to get it out and accessible.
I tested a bunch of the tant caps and they mostly look bad-ish. I'll post some examples soon. I couldn't exactly figure out how the capacitor may have failed so I downloaded LTSpice (scary! but Free!) and fumbled through making a schematic using a resistor and a non-ideal capacitor, a voltage source and ran a frequency response simulation...took a while to figure out how to do this. Then I right-clicked on the cap and changed the non-ideal values. Well, gimme a break, we were just watching Vera reruns. The upshot, when a capacitor ESR (equivalent series resistance) increases, ALOT (like maybe to 100 or 1000 ohms!) then the frequency plots start to look a lot like the ones we get on these capacitors.
I look up the failure mode of these tant's for the case when they show high ESR. It seems it can be caused by peak currents on startup or AC ripple. So then I realize there is a diode on the datasheet and not one in the turntable circuits. Then I remember reading it is really bad to reverse bias these ELECTROLYTIC tant caps. My guess, this is some kind of design problem with something happening at either power-up or power-down and a diode would have prevented this. It's a theory but all these caps going soft? Addendum: perhaps design problem is a little harsh...a design problem to make it last 40+ years.
Looks like a recap is coming. I may try something temporary to assure myself this will fix things before I take the main board out. Looks like it will be a major PITA to get it out and accessible.
Last edited:
mootej
Active Member
As promised, the rest of the tant cap testing attached as a pdf below. Quick look, I think C10, C16, C54 and C55 are OK. C22, C23, C26, C27, C28, C29, C30, C31, and C32 all look questionable, especially the phase plot.
Here was the LTSpice simulation of the same circuit with high ESR (100ohm):
And same circuit with a reasonable ESR (0.1ohm):
Sorry the phase plot is a little tough to see but I think you can see enough to get the idea that increasing ESR (alot) and the plots start to look like the caps in the Q11.
A few good caps and mostly bad. Here comes the onion, one more peel. What do the bad caps have in common? Except for C22 (argghh), the +5V power supply. Maybe a reverse bias transient or something? Up next, same Bat time, same Bat channel... let's look at the voltage on the cap and the +5V power supply at power-up and power-down.
Here was the LTSpice simulation of the same circuit with high ESR (100ohm):
And same circuit with a reasonable ESR (0.1ohm):
Sorry the phase plot is a little tough to see but I think you can see enough to get the idea that increasing ESR (alot) and the plots start to look like the caps in the Q11.
A few good caps and mostly bad. Here comes the onion, one more peel. What do the bad caps have in common? Except for C22 (argghh), the +5V power supply. Maybe a reverse bias transient or something? Up next, same Bat time, same Bat channel... let's look at the voltage on the cap and the +5V power supply at power-up and power-down.
Last edited:
mootej
Active Member
The continuing saga... the +5V power supply.
I measure two things on the oscilloscope, the output from the +5V power supply to the multivib's and the voltage across the tantalum caps attached to the multivib's. These set the length of the pulse from the multivib to the MPU. All of these caps have gone soft, not exactly blown but not healthy. Because of that, the E and F sensor do not work. In other words, the track select feature does not work.
I suspect the +5V power supply is the root-cause. Why would all the tant's blow? I suspect the power supply is biasing the electrolytic tant caps in reverse, at least enough to prematurely age them and make them all bad in the multivibrator circuits. The most likely time for this to happen is during power-up and power-down.
I measure two things on the oscilloscope, the output from the +5V power supply to the multivib's and the voltage across the tantalum caps attached to the multivib's. These set the length of the pulse from the multivib to the MPU. All of these caps have gone soft, not exactly blown but not healthy. Because of that, the E and F sensor do not work. In other words, the track select feature does not work.
I suspect the +5V power supply is the root-cause. Why would all the tant's blow? I suspect the power supply is biasing the electrolytic tant caps in reverse, at least enough to prematurely age them and make them all bad in the multivibrator circuits. The most likely time for this to happen is during power-up and power-down.
mootej
Active Member
[Oh, quick note, the times in these plots do not match up because I triggered the scope at different times in order to catch the events from the overall run in slow-motion. This little $50 scope is really limited in what it can do so I have to play games to get what I want.]
...but power-up is way more interesting than power-down. The long-run view:
The earlier part of the curve (the part of the curve just after 1 sec, above), with the time axis expanded to see the rise of the voltage on the capacitor:
This is crazy, the early part of the curve, expanded time scale to see the rise time on the +5V supply (above and expanded below):
Now what's crazy is the voltage is dipping to -0.6V for about 1 second during start-up. And the capacitor is seeing that negative bias. That's enough to fry all of those tant's, not right away but eventually. I attached a nice pdf with an overview of all the above power-up plots, showing the arrangement.
The big question, is this normal operation (i.e. we believe the Sansui engineers missed that one) or has something else, like a big electrolytic filter cap failed on the big power supply that supplies voltage +5V regulator?
Wow, this really is onion...
...but power-up is way more interesting than power-down. The long-run view:
The earlier part of the curve (the part of the curve just after 1 sec, above), with the time axis expanded to see the rise of the voltage on the capacitor:
This is crazy, the early part of the curve, expanded time scale to see the rise time on the +5V supply (above and expanded below):
Now what's crazy is the voltage is dipping to -0.6V for about 1 second during start-up. And the capacitor is seeing that negative bias. That's enough to fry all of those tant's, not right away but eventually. I attached a nice pdf with an overview of all the above power-up plots, showing the arrangement.
The big question, is this normal operation (i.e. we believe the Sansui engineers missed that one) or has something else, like a big electrolytic filter cap failed on the big power supply that supplies voltage +5V regulator?
Wow, this really is onion...
Last edited: