The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Gear for sale  Search Gearslutz Go Advanced
Dimension D Noise, Help!!
Old 4 days ago
  #1
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Dimension D Noise, Help!!

I've been killing myself for the past few days trying to sort out this issue. I'm getting a very jittery, crackly noise from both channels on my Dimension D, BUUUuuT only in certain conditions.

It took me a while to recreate the issue. Sine waves didn't induce the noise but I've found that sending a needle pulse tone above 1.3khz AND somewhere between -6 and -12db I can reproduce the noise (weird I know). If the tone is louder or softer than that range or lower in pitch, everything sounds fine.

I've recapped most of the unit, and socketed and replaced most of the ICs. I think I've narrowed it down to the inverting input on the NE5070N compander (pin 5). If I disconnect pin 5 the noisy range rises to about -4 to 0db , so something is definitely going on there.

Pleeeease someone give me some idea of what I might be missing. I'm all out of ideas.
Attached Thumbnails
Dimension D Noise, Help!!-screen-shot-2017-03-17-10.19.31-pm.png  
Old 4 days ago
  #2
Gear Maniac
 
rfnoise's Avatar
 

NE570n is a compander. Depending on how its internal sections are interconnected it functions as either a 2:1 compressor or 2:1 expander. The schematic you posted shows it is connected as a compressor. With active gain staging, weird things can happen. I am wondering if you have found a hiccup in the circuit design. Or perhaps the compander chip has developed an internal problem. Or, the compression is simply revealing a problem that is already present in the signal path before it. From my vantage point it is hard to know which.

The compander chip contains two independent channels. Your schematic shows one of them. Does the other channel use the other half of the chip, or does it use a second compander chip?

Pin 5 of the NE570n is the negative input to the chip's output opamp, and is used in the chip's output circuit feedback network. Pin 6 is the opamp's positive input.If you lift pin5 you break a feedback network for the output opamp. It will run with full gain, which is a lot, and thus any noise and other signal appearing on either pins 5 or 6 (or internally generated) will increase dramatically. It does not by itself indicate the chip is flawed.

Compressors/expanders have attack and release times, which are time constants that determine how quickly the amount of gain will change. In this circuit there is one capacitor that affects both. Your pulsed signal is probably of just the right intensity and timing to make the amount of gain change flop back and forth in such a way as to reveal an artifact of some kind that is otherwise masked. Since it is in both channels I am thinking it is not a defect in this single chip even if both channels are passing through the one NE570n. Because the channels are mostly independent of each other.

I have two thoughts about the next step to take. First, the non-compressed signal is applied to pin 6, coming from IC104's pin 1 (through C111). I wonder if you can monitor that signal by hooking it to a line input. Careful, it will have DC on it so you'd need to include a series capacitor to block dc into your input, using same polarity scheme as C111. If the artifact is present there, the problem is not the NE570n.

Second, capacitor C112 (.22uf) determines the compressor's attach and release times, MOSTLY release. It would be interesting to double its value and see what effect it has on the artifacts relative to your pulsing signal. Doubling its value is quick and easy to do by tack soldering another cap across it. I would expect it will change when/how the crackly noise appears. This would confirm that what you are hearing is somehow related to the compressor's action.
Old 4 days ago
  #3
Lives for gear
 
JohnRoberts's Avatar
 

IIRC that uses a BBD analog shift register for delay?

Is the noise in the dry path too, or only the delay? If just the delay it could be the BBD or the companding noise reduction.

A leaky cap in the compander circuit could introduce noise. That cap could be any one of the several associated with the compander (audio path, rectifier input, and even the rectifier output smoothing cap).

To debug whether it is compander or BBD, you need to isolate the BBD from the compress/expand circuitry.

JR
Old 3 days ago
  #4
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Quote:
Originally Posted by rfnoise View Post
NE570n is a compander. Depending on how its internal sections are interconnected it functions as either a 2:1 compressor or 2:1 expander. The schematic you posted shows it is connected as a compressor. With active gain staging, weird things can happen. I am wondering if you have found a hiccup in the circuit design. Or perhaps the compander chip has developed an internal problem. Or, the compression is simply revealing a problem that is already present in the signal path before it. From my vantage point it is hard to know which.

The compander chip contains two independent channels. Your schematic shows one of them. Does the other channel use the other half of the chip, or does it use a second compander chip?
Thanks for you're help.

The other half of the chip is being used as an expander. There is a second chip being used for the other channel. I tried swapping the chip with another working unit and it still had the same issue. Also both channels are showing the same weird behavior, so I don't think its the chip itself (I may be wrong).

I tried strapping a parallel .1uf cap across the .22uf on pin 1, but it didn't seem to have much effect. I had also previously swapped the .22uf caps between the compressor and expander halves of the chip, and the noise was still present on the compressor side.

Checking the input on pin 6 with a scope shows a clean signal, so the noise must be happen somewhere in the compressor circuit. I'm really hoping this isn't something inherent to the design, someone would have noticed this by now

Last edited by ltjohnrambo; 3 days ago at 11:42 PM..
Old 3 days ago
  #5
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Quote:
Originally Posted by JohnRoberts View Post
IIRC that uses a BBD analog shift register for delay?

Is the noise in the dry path too, or only the delay? If just the delay it could be the BBD or the companding noise reduction.

A leaky cap in the compander circuit could introduce noise. That cap could be any one of the several associated with the compander (audio path, rectifier input, and even the rectifier output smoothing cap).

To debug whether it is compander or BBD, you need to isolate the BBD from the compress/expand circuitry.

JR
I had previously tried swapping out the MN3007 BBD, but with no luck in getting rid of the noise.
Old 3 days ago
  #6
Gear Maniac
 
rfnoise's Avatar
 

Are you saying that you can SEE the noise on the scope on the output, but not on the signal going to pin 6? If you can see it, it is very very strong! Please confirm that you can see the noise!

John's suggestion to somehow remove the compander from this "equation" is a good one. Personally speaking, since this is a problem in both channels and the compander chips are different for each channel (if I understand you correctly), a bad chip is no longer possible. I see only three sources common to both channels.

1. Power supply
2. Noise inherent or imposed on the source
3. Artifacts of design

Have I missed a possibility?
Old 3 days ago
  #7
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Yeah, I can see jitter in the waveform coming off of pin 7. I had wrote earlier that I saw it present on pin 6, but I was mistaken.

I'll try to upload a video so you guys can see/hear what I mean.

I don't think its a power supply issue, since it only affects a certain type of waveform (something like a needle pulse or saw) at certain levels. I also recapped the whole powersupply.

The more I test, it does seem like it has something to do with the attack/release times on the compressor half of the NE570. Looking through the datasheet I'm not sure how the cap value on pin 1 affects the times. Would a larger value increase or decrease the time constants? These caps are also some of the only tantalum caps on the board, which makes me think maybe they all could have failed?
Old 3 days ago
  #8
Gear Maniac
 
rfnoise's Avatar
 

Quote:
These caps are also some of the only tantalum caps on the board, which makes me think maybe they all could have failed
Tantalum caps tend to far outlast electrolytics but the schematic does not indicate a polarized cap on pin 1. At .22uf and less than 50 volts, I don't see why they would use a polarized cap. Any type of cap would work as well. You could change them to try, but I do not expect an improvement.

Quote:
I'm not sure how the cap value on pin 1 affects the times. Would a larger value increase or decrease the time constants?
Lowering the cap's value will decrease attack and release times, increasing the value has the opposite effect. imagine the flow of audio is like a flow of water into a little tank of water with a hole in the bottom. Water (audio) flows into the tank from a pipe sometimes as drips, sometimes as little gushes or strong gushes, and when the overall flow is fast enough (loud enough) the water flows in faster than the drain hole can let it out. The size of the tank, the size of the pipe feeding in, and the size of the drain at the bottom affects how quickly the tank's water level can rise or fall. So, increasing the tank's size will reduce the speed the tank can be filled or emptied without changing the overall amount of water flowing. Increasing/decreasing that tank's size is like changing the value of our capacitor. Our "attack time" is like filling that tank, and our "release time" is like draining the tank

Taking this water tank analogy further, installing a float valve to sense the water level and restrict the flow of incoming water automatically is similar to an audio compressor's action.

Anyhow, back to the problem at hand. I look forward to your movie.
Old 3 days ago
  #9
Gear Maniac
 
rfnoise's Avatar
 

Quote:
The more I test, it does seem like it has something to do with the attack/release times on the compressor half of the NE570.
If I understand the NE570's data sheet, you can disable compander action by temporarily lifting pins 2 and 15.
Old 3 days ago
  #10
Lives for gear
 
JohnRoberts's Avatar
 

Quote:
Originally Posted by ltjohnrambo View Post
Yeah, I can see jitter in the waveform coming off of pin 7. I had wrote earlier that I saw it present on pin 6, but I was mistaken.

I'll try to upload a video so you guys can see/hear what I mean.

I don't think its a power supply issue, since it only affects a certain type of waveform (something like a needle pulse or saw) at certain levels. I also recapped the whole powersupply.

The more I test, it does seem like it has something to do with the attack/release times on the compressor half of the NE570. Looking through the datasheet I'm not sure how the cap value on pin 1 affects the times. Would a larger value increase or decrease the time constants? These caps are also some of the only tantalum caps on the board, which makes me think maybe they all could have failed?
The caps at pin 1 and 16(?) are indeed time constant or rectifier smoothing. Making them larger will slow both the attack and release.

I have never used tantalum there, aluminum seems just fine. make sure you use the same value for the compressor side and expander side or they will mistrack.

As I think I posted before any cap associated with the ne570 could cause noise if leaking.

That symptom also sounds like a bad solder connection so inspect for that too.

good luck

JR
Old 2 days ago
  #11
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Quote:
Originally Posted by JohnRoberts View Post
I have never used tantalum there, aluminum seems just fine. make sure you use the same value for the compressor side and expander side or they will mistrack.
I've got a .33uf ceramic and a 1uf electrolytic that I'll try out. What values are you normally using?

Also, thank you guys for all your help!
Old 2 days ago
  #12
Gear Maniac
 
rfnoise's Avatar
 

It is smart to rule out those caps by substitution, I'd temporarily sub in the .33's. Let's not forget though that we have an identical problem on two channels, which use two entirely different sets of components. Chances of failed parts on both channels in exactly the same way is small, not impossible, but small.

Does it not make sense to disable the companding action, as a test? Either the problem will go away or it won't, which will tell us a lot. I'll bet that you can scratch through a trace to pins 2 and 15 easily enough. Fixing traces is trivial, and doing a test that way will tell us very much.
Old 2 days ago
  #13
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Good idea, I'll try disengaging pins 2/15 too.
Old 2 days ago
  #14
Lives for gear
 
JohnRoberts's Avatar
 

anything in .47uF to 1uF should be OK...

JR
Old 2 days ago
  #15
For smaller values like 1 uf or less use a 5 mm film cap. For 2.2 uf or larger use a low leakage electrolytic. I used those NE570 compander chips in a video game PCM design I did for Hasbro back in the early 1990's. It used the 16 lines between video frames to encode it, you know, the black bars when a video screen rolls.

I also used one in a variable compressor/expander circuit using one knob, squish or expand. It was used in a spring reverb to adjust reverb time. I also used it on a chamber, a converted 1962 era bomb shelter with AKG 452 mics and a JBL 4311 speaker. Worked great.
Old 2 days ago
  #16
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter


Here's the vid, you can see the noise doesn't come in until I raise the volume on the tone.

I also tried disengaging pins 2 and 15 and that seems to make the noise louder.

EDIT: I increased the caps on pins 1 and 16 to 1uf, but that didn't seem to have any effect.
Old 2 days ago
  #17
Gear Maniac
 
rfnoise's Avatar
 

Thank you for the video! And thank you for going to the trouble of disabling the companding. I admit my unfamiliarity with this kind of effects unit has misled us a bit. JohnRoberts was going in the right direction already. The problem is being caused by whatever circuit is between the compressor (lowerhalf of the 570) and the expander (upper half). The noise is not present on pin 6 of that chip (the compressor's input) , but it will be present on the expander's input pin 12 and output pin 10. I would personally want to leave the compander disabled until the actual problem is found, but no harm either way.

So, now we're looking at the bucket brigade and anything else between compressor/expander. The quality of noise is very suggestive to me of digital clocking trouble or, perhaps, aliasing. If it is aliasing, though, I wouldn't expect it to be level sensitive like that. I wonder of audio is somehow leaking into the chip's clocking signal, perhaps a failed isolation cap? Can you post more of the schematic or give us a link to one?

I am still puzzled why both channels behave identically badly.
Old 2 days ago
  #18
Lives for gear
 
JohnRoberts's Avatar
 

Quote:
Originally Posted by ltjohnrambo View Post
Good idea, I'll try disengaging pins 2/15 too.
??? Disconnecting pins 2 and 15, (the rectifier inputs), will cause the compressor stage open up to maximum gain, and turn down the expander to full off... I don't expect that to reveal anything useful other than if the 570 is (mostly) working.

You seem to be approaching this like a car mechanic replacing parts. You have a scope, follow the audio path from input to output (with a sine wave input) and see where the audio path is being corrupted.

JR
Old 2 days ago
  #19
Gear Maniac
 
rfnoise's Avatar
 

Quote:
Originally Posted by JohnRoberts View Post
??? Disconnecting pins 2 and 15, (the rectifier inputs), will cause the compressor stage open up to maximum gain, and turn down the expander to full off.
No that's not true. I think you have confused the rectifier input with the inverse input on pin 6. If you review the data sheet I am sure you will see that with nothing connected on pin 2/15, the rectifier input will remain at idle. It will remain inherently DC stable and also generate a zero AC signal condition, so the gain stage it controls will neither compress nor expand. It is safe to leave disconnected for this test.

When the OP disconnected those pins the noise got louder, BECAUSE THERE IS NO COMPANDING ACTION. That is exactly what one would expect to happen.

edit: I re-read your post and I thought you were saying the gain would open up to full gain. I mis-read. You are correct in that, by disconnecting those pins, at my request, the OP disabled the 570's action, so it became a simple signal pass-through. Yes, the idea was to prove that the noise is not a result of gain control. We can now stop looking at the 570 as the cause of the problem.
Old 2 days ago
  #20
Lives for gear
 
JohnRoberts's Avatar
 

Quote:
Originally Posted by rfnoise View Post
No that's not true. I think you have confused the rectifier input with the inverse input on pin 6. If you review the data sheet I am sure you will see that with nothing connected on pin 2/15, the rectifier input will remain at idle. It will remain inherently DC stable and also generate a zero AC signal condition, so the gain stage it controls will neither compress nor expand. It is safe to leave disconnected for this test.
I would prefer to not argue about how the 570 works. I published a kit article for a tape noise reduction using it in 1977. Since then I have done numerous designs with 570 or variants (like 572).

The 570 uses a current in/current out transconductance gain cell that is modulated by the bias current rectified from the AC signal at pin 2 (15). There is an internal 10k resistor that converts the pin 2 signal to current before rectification, and smoothing by cap at pin 1. The gain cell has an internal 20k resistor to convert voltage to current for the current ratioing gain cell.

IIRC unity gain is something like 140 uA of gain cell bias current (might be 280 uA). No bias current makes the the gain cell appear like a very high impedance. For the compressor the gain cell is in the feedback path so no bias commands high voltage gain, for the expander the gain cell is in the forward path so no bias current commands low/no output.

The wide open compressor will boost input noise tens of dB but it is unclear what the OP is monitoring, since the expandor output should be symmetrically reducing gain with the rectifier open.

If pin 2 is open but pin 15 still connected it will be noisy as hell.
Quote:
When the OP disconnected those pins the noise got louder, BECAUSE THERE IS NO COMPANDING ACTION. That is exactly what one would expect to happen.
no... but perhaps compression without the decoding expansion, if pin 15 still connected.
Quote:
edit: I re-read your post and I thought you were saying the gain would open up to full gain. I mis-read. You are correct in that, by disconnecting those pins, at my request, the OP disabled the 570's action, so it became a simple signal pass-through. Yes, the idea was to prove that the noise is not a result of gain control. We can now stop looking at the 570 as the cause of the problem.
To convert the 570 to a unity gain pass through you need to trick the rectifier inputs to think they have nominal 140uA audio signal.

While I do not recommend doing this, shorting both pins 2 and 15 to ground would send 180uA DC into the two rectifiers, so it should work as a slightly less than unity gain compressor, with slightly more than unity gain expander that would deliver unity gain over all.

I repeat I do not advise doing this. If the op has a scope... just look for where the noise is coming from.

JR

PS: Please don't lecture me about how chips work that I am intimately familiar with, it confuses the lurkers and scares the children.
Old 2 days ago
  #21
Gear Maniac
 
rfnoise's Avatar
 

What I suggested DOES increase the noise floor exponentially (something like 40db more, while the pins are disconnected) but that does not matter. It was to show that the problem is not caused by compander action. Presuming the OP did what he said he did, he disabled companding by isolating those two pins yet the problem noise persists. That's all we need to know to look elsewhere. The brigade is next.

To the OP, I can see JohnRoberts wants to be involved more closely and I encourage him to do so. I've done my best to keep the ball rolling for you.
Old 2 days ago
  #22
Gear Addict
 
ltjohnrambo's Avatar
 

Thread Starter
Let me try to clarify what I've tried so far.

Quote:
If pin 2 is open but pin 15 still connected it will be noisy as hell.
When I tested this I disconnected both 2 and 15. This resulted in a change to the noise that sounded like a secondary oscillation to the test tone. The secondary oscillation rose and fell in frequency when I raised and lowered the gain of the test tone. If it helps I can post a video of this.

Quote:
You have a scope, follow the audio path from input to output (with a sine wave input) and see where the audio path is being corrupted.
As far as I can tell the test tone is ok when entering the NE570 at pin 6, I've tested this with the scope and by tapping a line out to a speaker. The noise is then present on pin 7 output. This makes me think there must be something going on with the NE570 or with the immediately surrounding circuit. I've replaced most of the components in the immediate area, so I'm not sure what it could be. I've also swapped the NE570 with a working Roland SBF-325 with no improvement. It's possible that the SBF's 570 was also broken but not exhibiting the same behaviour due to differences in the circuit design
Old 1 day ago
  #23
Gear Maniac
 
rfnoise's Avatar
 

Okay, well, then I guess we're going to keep looking at the 570. Is the cap on pin 8 really grounded and does behavior change by increasing its value (anything non-polarized between 100pf and .1uf)? There is a 10p cap between pins 5 and 7. What happens if you tack another cap across it, anything from 10-50p temporarily? That should further reduce high freq. response and maybe tame an instability that is triggered by the harmonics-filled signal you are testing with.

If nothing above helps, what happens if you unhook one side of R118? (12k I think it says but my aged eyes are not certain). This disconnects all circuit beyond the 570. Is distortion still present on pin 7? (edit, yes this will upset the circuit immediately after it but shouldn't harm anything)

edit: keep leads of any new caps as short as possible. Don't want the leads to become antennas.
Old 1 day ago
  #24
Lives for gear
 
JohnRoberts's Avatar
 

Quote:
Originally Posted by ltjohnrambo View Post
Let me try to clarify what I've tried so far.


When I tested this I disconnected both 2 and 15. This resulted in a change to the noise that sounded like a secondary oscillation to the test tone. The secondary oscillation rose and fell in frequency when I raised and lowered the gain of the test tone. If it helps I can post a video of this.
not for my benefit... I still advise against doing that.
Quote:

As far as I can tell the test tone is ok when entering the NE570 at pin 6, I've tested this with the scope and by tapping a line out to a speaker.
pin 6 is input to compressor, clean is good.
Quote:
The noise is then present on pin 7 output. This makes me think there must be something going on with the NE570 or with the immediately surrounding circuit. I've replaced most of the components in the immediate area, so I'm not sure what it could be. I've also swapped the NE570 with a working Roland SBF-325 with no improvement. It's possible that the SBF's 570 was also broken but not exhibiting the same behaviour due to differences in the circuit design
Unlikely to find two identically bad ICs.

Is pin 2 connected or floating? If floating (stop doing that) the compressor will be running at very high gain. If pin 2 is connected signal at 7 should look similar to signal at 6 but with only modest level change.

As i've suggested before pretty much every cap connected to the 570 could cause noise if leaky.

JR
Loading mentioned products ...
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get instant access to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump