Author Topic: SAD1024 clock noise help  (Read 4364 times)

anotherjim

Re: SAD1024 clock noise help
« Reply #60 on: October 26, 2022, 04:31:35 AM »
I find most transistors can be checked in place with DMM diode check with power off. Every base is a diode to its emitter and collector and should only read in one polarity while collectors don't read to the emitter either way. When you find one breaks the rules, look at the schematic and see if the circuit has other things connected that account for it.

With a fragile PCB, make sure you have a replacement part and clip the leads/pins away with flush cutters. that makes it easier to get the remaining stub out of the PCB.
If you think one IC pin is shorting out, clip just that pin top side and see if that stops the short. If that pin turns out to be innocent, you can reattach it with solder or a bit of wire.

puretube

Re: SAD1024 clock noise help
« Reply #61 on: October 26, 2022, 04:46:19 AM »
You may try this: cut the PCB-trace from pin13 of IC7 (the non-working Q-output of the 4013). Hook up an inverting logic gate (whatever you got) input to pin12 of IC7. On the output of that gate you now have the same logic clock-state you normally would get from pin13 of IC7. Connect this output of the gate to the end of the cut trace that goes to the SAD1024 (where normally pin13 of IC7 would have gone).
Don`t forget the power-supply for the inverting-gate-chip.
If the testpoint 7 now still shows "hash" instead of an inverted clock compared to testpoint 6: OUCH - the BBD is very probable seriously hurt. (Possibly from receiving clock-pulses at the time when the broken-down transistor Q20 had cut off the main power-supply to the BBD?).

four_corners

Re: SAD1024 clock noise help
« Reply #62 on: October 26, 2022, 11:51:39 AM »
You may try this: cut the PCB-trace from pin13 of IC7 (the non-working Q-output of the 4013). Hook up an inverting logic gate (whatever you got) input to pin12 of IC7. On the output of that gate you now have the same logic clock-state you normally would get from pin13 of IC7. Connect this output of the gate to the end of the cut trace that goes to the SAD1024 (where normally pin13 of IC7 would have gone).
Don`t forget the power-supply for the inverting-gate-chip.
If the testpoint 7 now still shows "hash" instead of an inverted clock compared to testpoint 6: OUCH - the BBD is very probable seriously hurt. (Possibly from receiving clock-pulses at the time when the broken-down transistor Q20 had cut off the main power-supply to the BBD?).

Ah thats a good idea. I can also just pull pin13 of IC7 out of the socket instead of cutting the trace I'm guessing, so I might just do that.

I have a 4011 I can try using, but since it is a NAND gate, I can essentially just send pin12 of IC7 to the 4011, and short two of the inputs and it should output the inverted clock, correct?

puretube

Re: SAD1024 clock noise help
« Reply #63 on: October 26, 2022, 06:40:42 PM »
All unused inputs of possible other gates of the "help-chip" should be grounded.

four_corners

Re: SAD1024 clock noise help
« Reply #64 on: October 26, 2022, 08:08:53 PM »
All unused inputs of possible other gates of the "help-chip" should be grounded.

Well, in the strangest set of events, I figured out the issue with the effect (spoiler: everything works again!).

After going through all the suggestions, I went to close up the synth to deal with it another day, and I hear a tiny metal sound. Turns out, a component lead fell out, and the effect worked again. I guess a loose lead slid under some components, so I couldn't see it at all when checking the pcb for issues top to bottom.

_____

So, back to the bias!

I'm able to get the clock frequency at max Manual to the required 1MHz, but I cannot seem to get Manual at 0 anywhere close to 100kHz. I can get it to around 400kHz, but that also brings the width of the frequency very narrow (max ends up being around 600kHz).

I tried changing R90 from 430k to 200k, which allowed the Manual clock frequency to get down below 100kHz, but brought in other noisy stuff and the max Manual wouldn't get much above 500kHz.

Any ideas?

It seems like something in this region would mainly change the intensity and not the bias...



I guess then something in this region could be tweaked?


pacealot

Re: SAD1024 clock noise help
« Reply #65 on: October 27, 2022, 01:31:43 PM »
I'm not at all certain that this is relevant due to the topography being so different, but I recently had a similar issue with achieving the correct clock rates on an A/DA flanger clone, and that experience suggested to me that perhaps your C46, the 22pF cap off two terminals of the 4013, might be tweakable to help solve this issue. I'd love to hear more thoughts from the great BBD minds hereabouts.

(by the way, congrats on getting everything to work again I was lurking out here holding my breath for the health of your SAD1024 for a goodly while...)
"Unneeded bandwidth is the devil's playground" idy

anotherjim

Re: SAD1024 clock noise help
« Reply #66 on: October 27, 2022, 05:04:52 PM »
Quote
I was lurking out here holding my breath for the health of your SAD1024 for a goodly while...
As was I...
22pf cap on the 4013 is very much involved in the max clock frequency, but I doubt there is much room for tweaking. You might find an 18pf will go faster - if you can find one. When cap values get this small, stray and pin capacitance starts to dominate. Maybe 6pf there already and it might even run super-stupid fast with no cap at all. But then again, a 10pf together with the strays might speed up a bit and still work. There also the possibility of a faster chip with an MC14013 - would be interesting to see the difference.
There may be some room to reduce the minimum charge/discharge rate of the existing cap by tweaking some resistance values. I have no suggestion of what at the mo.
The danger with too fast is the clock wave shapes get too rounded and lose peak amplitude then the BBD stages can't do their bucket handover trick too well.

four_corners

Re: SAD1024 clock noise help
« Reply #67 on: October 27, 2022, 05:32:15 PM »
Quote
I was lurking out here holding my breath for the health of your SAD1024 for a goodly while...
As was I...
22pf cap on the 4013 is very much involved in the max clock frequency, but I doubt there is much room for tweaking. You might find an 18pf will go faster - if you can find one. When cap values get this small, stray and pin capacitance starts to dominate. Maybe 6pf there already and it might even run super-stupid fast with no cap at all. But then again, a 10pf together with the strays might speed up a bit and still work. There also the possibility of a faster chip with an MC14013 - would be interesting to see the difference.
There may be some room to reduce the minimum charge/discharge rate of the existing cap by tweaking some resistance values. I have no suggestion of what at the mo.
The danger with too fast is the clock wave shapes get too rounded and lose peak amplitude then the BBD stages can't do their bucket handover trick too well.

It seems like I need the opposite actually, not faster clock but slower, as I'm having trouble getting the bias anywhere close to the bottom end of the range (100 kHz).

On post #34, puretube mentioned a mistake on the calibration doc...


On page 22, point 8. of the service-manual:
https://www.vintagesynthparts.com/wp-content/uploads/2018/01/Korg_Trident_MK1_Service_Manual.pdf

BUT: it contains a typo IMHO!
It says the idle clock-frequency at the lowest "MANUAL"-setting (VR11) should be roughly 100kHz (90~110kHz).
Okay!
It further says the clock-frequency at the highest "MANUAL"-setting (VR11) should be roughly 1kHz (0.9~1.1kHz).
FAIL! (by a factor of thousand).
This correctly should read roughly 1MHz (0.9MHz~1.1MHz i.o.w.: 900kHz~1100kHz).
During the measuring, "INTENSITY" (VR10) needs to be turned waydown to zero!

« Last Edit: October 27, 2022, 05:34:05 PM by four_corners »

four_corners

Re: SAD1024 clock noise help
« Reply #68 on: October 27, 2022, 06:44:30 PM »
So regardless of what the manual says, I think I got the bias set correctly.

When I actually let it Flange with the clock at max Intensity, the range is from 100 kHz 1 mHz. At full Intensity, the Manual pot does nothing, but as varying levels of the Intensity pot, the Manual pot essentially scrubs to different areas of that 100 kHz to - 1 mHz range, and when Intensity is at zero (no flanging), the Manual pot lets me manually flange from about 600 kHz to roughly 1.1 mHz.

The reason I'm okay with this is because I'd rather have it sound good during actual clock modulated Flanging, rather than get the Manual pot correct and have the actual clock flanging sounding totally insane.
« Last Edit: October 27, 2022, 07:37:47 PM by four_corners »

Rob Strand

Re: SAD1024 clock noise help
« Reply #69 on: October 27, 2022, 07:14:46 PM »
Quote
BUT: it contains a typo IMHO!
It says the idle clock-frequency at the lowest "MANUAL"-setting (VR11) should be roughly 100kHz (90~110kHz).
Okay!
It further says the clock-frequency at the highest "MANUAL"-setting (VR11) should be roughly 1kHz (0.9~1.1kHz).
FAIL! (by a factor of thousand).
This correctly should read roughly 1MHz (0.9MHz~1.1MHz i.o.w.: 900kHz~1100kHz).
During the measuring, "INTENSITY" (VR10) needs to be turned waydown to zero!
Definitely a typo.   

1kHz clock doesn't make any sense *at all*.

A 10:1 range on the clock (100kHz to 1MHz) for the Manual control seems very reasonable.
Send:     . .- .-. - .... / - --- / --. --- .-. -

four_corners

Re: SAD1024 clock noise help
« Reply #70 on: October 27, 2022, 07:37:22 PM »
Quote
BUT: it contains a typo IMHO!
It says the idle clock-frequency at the lowest "MANUAL"-setting (VR11) should be roughly 100kHz (90~110kHz).
Okay!
It further says the clock-frequency at the highest "MANUAL"-setting (VR11) should be roughly 1kHz (0.9~1.1kHz).
FAIL! (by a factor of thousand).
This correctly should read roughly 1MHz (0.9MHz~1.1MHz i.o.w.: 900kHz~1100kHz).
During the measuring, "INTENSITY" (VR10) needs to be turned waydown to zero!
Definitely a typo.   

1kHz clock doesn't make any sense *at all*.

A 10:1 range on the clock (100kHz to 1MHz) for the Manual control seems very reasonable.

I'm only getting 600 kHz to roughly 1.1 mHz for the Manual control range, but when I max the Intensity control, I get a range of 100kHz to 1MHz, so I'm just gonna stick with that. It sounds awesome, so I'm going by that, haha.

As far as the ticking, here is my scope out of the SAD1024 but before the audio output of the board (I took one of these at TP8 directly out of the BBD, and one at the Feedback pot, but I can't remember which image is which). The synth was set to sawtooth, hence the shape.





I'm guessing the spikes are the clock ticking?

Any sort of capacitor I could connect to pin12 for the BBD to smooth this?
« Last Edit: October 27, 2022, 07:51:31 PM by four_corners »

Rob Strand

Re: SAD1024 clock noise help
« Reply #71 on: October 27, 2022, 07:51:52 PM »
Quote
We are getting somewhere!!! So the cap from ground to Manual pot wiper didn't fix anything... BUT, I tried hooking the cap to one side of the Intensity pot, and the clicking totally turns from clicking, to a smoothed static sound!!! With Intensity at 10, there is a new sort of high resonance heart beating sound, but I tried a green 470nF polyester cap I had sitting next to me and the heart beating sound dropped down a ton. This lug on the Intensity pot is connected to Pin 1 of IC12.

If I want to try out a higher value electrolytic, which direction would I place it? Positive lead to pin 1 and then negative to ground?

(EDITED per next page correction)

The whole idea of R102, C50, and IC12 is to filter the LFO output so it's a bit weird that putting caps on the intensity pot has an effect.   It seems like it's hiding another problem.

I have a suspicion something weird is going on around IC12.  The thing that stands out to me is R107.   Perhaps try adding a 100nF cap across R107, that will prevent noise getting onto pin 2.
« Last Edit: October 27, 2022, 08:18:36 PM by Rob Strand »
Send:     . .- .-. - .... / - --- / --. --- .-. -

Rob Strand

Re: SAD1024 clock noise help
« Reply #72 on: October 27, 2022, 08:02:58 PM »
Quote
I'm only getting 600 kHz to roughly 1.1 mHz for the Manual control range, but when I max the Intensity control, I get a range of 100kHz to 1MHz, so I'm just gonna stick with that. It sounds awesome, so I'm going by that, haha.

You might have to decrease VR4, then after that tweak VR3.   The idea is to find the magic combination of VR3 and VR4 that gives you the correct frequency at each end of the Manual control.   It might help to work out which pot has more effect on which end although some playing will generally get you there.

Quote
I'm guessing the spikes are the clock ticking?

Any sort of capacitor I could connect to pin12 for the BBD to smooth this?
It's doubtful.    High frequency circuit have all sorts of glitches.

The thing is the glitches you are seeing are very fast but the ticks you are hearing are synchronized with the LFO.  It is more likely the ticks originate from the LFO.   Audible glitches in audio might not even be visible on the oscilloscope.   One way to look for then is to trigger the oscilloscope off the LFO then probe with the other CRO channel.   All you would be doing is displaying what your ear already hears.

It's hard to get slow glitches from high frequency circuit.   You need two clocks which are nearly equal.
« Last Edit: October 27, 2022, 08:04:40 PM by Rob Strand »
Send:     . .- .-. - .... / - --- / --. --- .-. -

four_corners

Re: SAD1024 clock noise help
« Reply #73 on: October 27, 2022, 08:27:32 PM »
Quote
We are getting somewhere!!! So the cap from ground to Manual pot wiper didn't fix anything... BUT, I tried hooking the cap to one side of the Intensity pot, and the clicking totally turns from clicking, to a smoothed static sound!!! With Intensity at 10, there is a new sort of high resonance heart beating sound, but I tried a green 470nF polyester cap I had sitting next to me and the heart beating sound dropped down a ton. This lug on the Intensity pot is connected to Pin 1 of IC12.

If I want to try out a higher value electrolytic, which direction would I place it? Positive lead to pin 1 and then negative to ground?

(EDITED per next page correction)

The whole idea of R102, C50, and IC12 is to filter the LFO output so it's a bit weird that putting caps on the intensity pot has an effect.   It seems like it's hiding another problem.

I have a suspicion something weird is going on around IC12.  The thing that stands out to me is R107.   Perhaps try adding a 100nF cap across R107, that will prevent noise getting onto pin 2.

If I'm getting a clean sine wave out of pin1 on IC12, could the ticking still be hidden within a clean looking sine wave? Regardless, I might as well try a cap across R107.

Rob Strand

Re: SAD1024 clock noise help
« Reply #74 on: October 27, 2022, 09:23:27 PM »
Quote
If I'm getting a clean sine wave out of pin1 on IC12, could the ticking still be hidden within a clean looking sine wave? Regardless, I might as well try a cap across R107.
If it looks clean it's probably unlikely to be the cause.   I guess that brings up the question: what is adding the cap to IC12 / the intensity pot actually doing?   It's not clear what's going on.   So maybe try the cap across R107 but don't expect much.

If the cause of the tick is in fact the LFO then adding the RC network that you posted back in reply #21 is very much worthwhile.   Again no guarantees.

There's other slightly off the norm stuff in the circuit like R105 through IC12b, which tweaks the BBD bias depending on the clock frequency.   They might be going a bit overboard here. That has the potential to inject junk from the LFO into the audio path.   The idea is C51 filters any junk coming through from the LFO/Manual Ckt ie. Korg have thought about it. 

You could take a big step back and kill the LFO by putting a short across C49.  That will stop the LFO oscillating.  If the tick is still present it's not the LFO,  well not the Flanger LFO, as there's others in the circuit.
Send:     . .- .-. - .... / - --- / --. --- .-. -

four_corners

Re: SAD1024 clock noise help
« Reply #75 on: October 27, 2022, 09:37:35 PM »
Quote
If I'm getting a clean sine wave out of pin1 on IC12, could the ticking still be hidden within a clean looking sine wave? Regardless, I might as well try a cap across R107.
If it looks clean it's probably unlikely to be the cause.   I guess that brings up the question: what is adding the cap to IC12 / the intensity pot actually doing?   It's not clear what's going on.   So maybe try the cap across R107 but don't expect much.

If the cause of the tick is in fact the LFO then adding the RC network that you posted back in reply #21 is very much worthwhile.   Again no guarantees.

There's other slightly off the norm stuff in the circuit like R105 through IC12b, which tweaks the BBD bias depending on the clock frequency.   They might be going a bit overboard here. That has the potential to inject junk from the LFO into the audio path.   The idea is C51 filters any junk coming through from the LFO/Manual Ckt ie. Korg have thought about it. 

You could take a big step back and kill the LFO by putting a short across C49.  That will stop the LFO oscillating.  If the tick is still present it's not the LFO,  well not the Flanger LFO, as there's others in the circuit.

If I short IC11 across C49, will that potentially hurt anything? I guess I'm asking, when you say "kill" the LFO, are you saying kill as in fry, or just the LFO stopping as the inverted input and output will be shorted.

Rob Strand

Re: SAD1024 clock noise help
« Reply #76 on: October 27, 2022, 09:41:50 PM »
Quote
If I short IC11 across C49, will that potentially hurt anything? I guess I'm asking, when you say "kill" the LFO, are you saying kill as in fry, or just the LFO stopping as the inverted input and output will be shorted.
It should be fine.   Just make sure it *is* actually C49.  It doesn't hurt to check value and that the tracks go to the right pins on IC11 (sometimes schematics don't match actual boards).

Shorting C49 will stop the LFO from oscillating.   If you still hear a tick it can't possibly come from the LFO.   You can look at the LFO output on the oscilloscope to be doubly sure the oscillator has stopped.
Send:     . .- .-. - .... / - --- / --. --- .-. -

four_corners

Re: SAD1024 clock noise help
« Reply #77 on: October 27, 2022, 11:38:41 PM »

You might have to decrease VR4, then after that tweak VR3.   The idea is to find the magic combination of VR3 and VR4 that gives you the correct frequency at each end of the Manual control.   It might help to work out which pot has more effect on which end although some playing will generally get you there.


To quickly jump back to the bias stuff, is there any way you might be able to explain how the Manual pot works? It looks like it is mixing both the +15V and -15V in some way. Seems to make sense that the Manual pot would do nothing with Intensity pot at 10 as it would allow the sine wave at IC12 pin1 to flow through and cut off the voltage from the Manual pot.

Basically just wondering what changing the resistance of either the Manual pot, or the 100K resistor right after it would do. With Intensity at 10, the frequency range is perfect (100kHz to 1mHz), but it would be nice to be able and have the Manual pot range match when Intensity is at 0. If I try to get VR3 and VR4 close to the "correct" range, the sweep of the actual full Intensity flange is crazy and sounds insane, the range goes down super low and super high.



Anyway, it sounds great right now, its just the Manual range with Intensity at 0 is about 400-600kHz. Not a huge deal to be honest. If I'm not able to fix the clicking, I can always count on my trusty Ibanez NB-10 Noise Buster, which is essentially just a filter and not a noise gate.

four_corners

Re: SAD1024 clock noise help
« Reply #78 on: October 28, 2022, 12:40:02 AM »
Quote
If I short IC11 across C49, will that potentially hurt anything? I guess I'm asking, when you say "kill" the LFO, are you saying kill as in fry, or just the LFO stopping as the inverted input and output will be shorted.
It should be fine.   Just make sure it *is* actually C49.  It doesn't hurt to check value and that the tracks go to the right pins on IC11 (sometimes schematics don't match actual boards).

Shorting C49 will stop the LFO from oscillating.   If you still hear a tick it can't possibly come from the LFO.   You can look at the LFO output on the oscilloscope to be doubly sure the oscillator has stopped.

Shorting C49 kills the LFO and kills the ticking, obviously this kills the flanging too. I tried adding a bigger cap here, and it just slows down the LFO frequency as I expected. Technically this doesn't rule out something further down the chain though right? As the LFO could be picking up the tick later on.

Maybe I'll try the Stomboxology rounded waveform mod again and actually monitor it on my scope like a smart person would have done. Last time I just tried it, still heard the ticking, and kept on trying other things. This is the RC network you mentioned, correct?

Would there be any use in adding a second-order filter for power? Essentially changing it from the left image to the right image?


Or even just making R94 and R95 larger than 100ohm, and C43 and C44 larger than 100uf?

Thanks again everyone for your help and suggestions. I hate to waste everyones time with such a small annoyance, but I really appreciate it.

Rob Strand

Re: SAD1024 clock noise help
« Reply #79 on: October 28, 2022, 03:15:04 AM »
Quote
To quickly jump back to the bias stuff, is there any way you might be able to explain how the Manual pot works? It looks like it is mixing both the +15V and -15V in some way. Seems to make sense that the Manual pot would do nothing with Intensity pot at 10 as it would allow the sine wave at IC12 pin1 to flow through and cut off the voltage from the Manual pot.
R106 and R108 divider the voltage.   That's pretty normal thing to do.  The idea being is the Manual control only needs to shift the delay a certain amount.   The swing of the LFO triangle wave is probably around +/-9V or +/-10V.   The manual control after R106/R108 is about +/- 9.6V.    So that pretty normal that the range of the Manual Control matches the LFO.   The Q21, Q22 circuit just buffers the reduced Manual voltage and produces the Manual voltage at Q22's emitter.

So the Depth pot simply mixes the LFO voltage and the Manual Voltage.   At low intensity the voltage at the Intensity wiper varies +/-9.6V (say) depending on where the Manual pot is set.  At full intensity it the voltage at the Intensity wiper varies +/-9.6V at the LFO rate.   In between settings let you set how much swing the LFO has and the Manual control moves that up or down a bit to set the middle of the delay.

All totally normal stuff.

Quote
Basically just wondering what changing the resistance of either the Manual pot, or the 100K resistor right after it would do. With Intensity at 10, the frequency range is perfect (100kHz to 1mHz), but it would be nice to be able and have the Manual pot range match when Intensity is at 0. If I try to get VR3 and VR4 close to the "correct" range, the sweep of the actual full Intensity flange is crazy and sounds insane, the range goes down super low and super high.
So it should already do that.

One thing be careful about is:  When you use the Manual control it is very easy to measure the clock.   As a result you can see that the manual control isn't reaching the limits in the Service Manual.   *However* what if the VCO or the VCO adjustments are the cause and not the Manual control circuit.  To me the Manual control circuit looks fine (You could verify the range of voltages at the emitter of Q22 is +/- 9.6V at the Extremes of the manual control.)   If the VCO isn't meeting spec with the Manual control then it won't be meeting the spec when the LFO is sweeping.   With Intensity on full it should sweep the full clock range as well (or close to it).   This is more difficult to measure so you are less likely to pick up a problem,  even though the same problem exists with both the Manual control and the LFO waveform.   That's also why it's not a good idea to"patch fix" the behaviour based on only what you see with the Manual control.   You probably need to dig further.   So first verify the voltage range Q22 is +/- 9.6V.   Then verify the LFO swings +/- 9.6V at IC12 pin 1 (it will actually swing a little less due to the filter.)

The RC network which can remove the ticks is R1 and C1 on this pic,

https://postlmg.cc/MMB6Rsdz

If the value R*C is too big it will slow down the LFO a biit.

Quote
Would there be any use in adding a second-order filter for power? Essentially changing it from the left image to the right image?
From what I can see normally +/- 15V rails come from the power supply and hit the two cap C48, C47 which are the filtering for the whole board.   The +/-15V rails then go to IC11.

So other than C48 and C47 there's not a lot of supply filtering.

In order to add filtering to IC11 (the LFO) you would need to replicate a circuit like R94, R95, C43, C44 then power IC11 from that "new" filtered power rail.

Like all these mods you have to try stuff to narrow down the problem.   Sometimes the things you try don't fix the problem ... because that doesn't turn out to be the problem.      There's no easy way to know without trying it.
« Last Edit: October 28, 2022, 03:19:00 AM by Rob Strand »
Send:     . .- .-. - .... / - --- / --. --- .-. -