THAT 1512 micpre noise issue

[user 133392]

I didn't read the entirety of this thread but it looks like on the layout you also have the 151X output, pin 6, leading to a connector that runs in parallel with pin 1's Rgain line. Those two are likely capacitive coupling resulting in peaking and "tail-biting" at high gain.



Avoid meandering Rgain lines and capacitive coupling of Rgain (pins 1 and 8) to other lines and/or ground.

Maybe I missed it on the layout but where is R8 the 10Ω gain limit resistor in series with the 1000 µF?
If missing you have no maximum gain limit stop resistor.

The grounding/layout routing is a mess to be completely honest.

 

[Mr. Moscode]

Nice and thank you all for these good willing infos and that you all suggest several general things instead of having experience with that THAT1512 circuit. Like suddenly changing r7 even if it´s suggested by the manufacturer and noone ever seems to have a problem with this in a lot of designs. Fyi when I reduced the overall gain of the circuit to around 55db everything is fine and dead quiet.
Only when I want to reach 60 db the ciruit behave that way.
But these general pin1 problem sounds a little like fishing in the dark for me. Unless you had the same project on the bench or built a 1512 pre
it´s not really that helpful. You never know if the chip behaves exactly how it does here even with the things changed.
" running some heavier gauge wire from the ground points at each amp to the main ground." Was the first I did years ago. Didn´t change anything.
I finished hundreds of projects and I am into electronics since a long time so no newbie here,
I know about star grounding and stuff but I also know that a lot of these generell suggestion often does not hit the target at all.
When this all should be true nobody ever would have finished any GSSL ever
I know that you test a micpre with a termination resistor but I can do further more testing the way I want and I know that when I have nothing plugged in and an obvious noise appears at one setting at my metering something is not as supposed to be compared to other micpres I have here.

I use a lorlin switch for the gain with an anti log resistor curve that the last resistor now has 10 ohm instead of 5 ohms which helps a lot and did nothing to any grounding or rail layout yet.
When I gain up the circuit now I have pure silent gain til the last step with a mic plugged in. When I unplugged anything I just have a little hiss at the highest settings. When I did that before I have terrible hiss at the last settings. So maybe the that chip without any servo or addition only is stable til 55db.
The jdk R20 uses the same chip and also only goes til 55db.
So maybe the answer is THAT simple

If the fader grounds an input, even through a 10 ohm R I would be looking for an oscillation. One way to stop that is to put some resistance in line with the input to the op amp. The give away is that the local amp input goes wild when close to ground. putting some R in series with the input will still let the fader go to 0 but the input doesn't see ground.
Ground trace inductance can turn a ground trace into a tank circuit of an oscillator.

 

[Mr. Moscode]

I didn't read the entirety of this thread but it looks like on the layout you also have the 151X output, pin 6, leading to a connector that runs in parallel with pin 1's Rgain line. Those two are likely capacitive coupling resulting in peaking and "tail-biting" at high gain.

View attachment 103003

Avoid meandering Rgain lines and capacitive coupling of Rgain (pins 1 and 8) to other lines and/or ground.

Maybe I missed it on the layout but where is R8 the 10Ω gain limit resistor in series with the 1000 µF?
If missing you have no maximum gain limit stop resistor.

The grounding/layout routing is a mess to be completely honest.

There are some too close traces there! A lot of stray C!

 

[Bo Deadly]

I didn't read the entirety of this thread but it looks like on the layout you also have the 151X output, pin 6, leading to a connector that runs in parallel with pin 1's Rgain line. Those two are likely capacitive coupling resulting in peaking and "tail-biting" at high gain.

The gain network is pretty low Z at high gain so I don't think it's an immediate source of instability.

However, it could easily be a source of noise. The return follows source rule is not being applied here. The gain control network lines should be closely parallel, use adjacent connector pins and the wires should be twisted to the pot.

The grounding/layout routing is a mess to be completely honest.

Ok, you hold him down while I kick him!

 

[user 133392]

"The gain network is pretty low Z at high gain so I don't think it's an immediate source of instability."

Oh really?

https://thatcorp.com/datashts/Analog_Secrets_Your_Mother_Never_Told_You.pdf pdf pages 16-18.

And I'll ask again: "Maybe I missed it on the layout but where is R8 the 10Ω gain limit resistor in series with the 1000 µF?
If missing you have no maximum gain limit stop resistor."

 

[timpanic]

"The gain network is pretty low Z at high gain so I don't think it's an immediate source of instability."

Oh really?

https://thatcorp.com/datashts/Analog_Secrets_Your_Mother_Never_Told_You.pdf pdf pages 16-18.

And I'll ask again: "Maybe I missed it on the layout but where is R8 the 10Ω gain limit resistor in series with the 1000 µF?
If missing you have no maximum gain limit stop resistor."

Thank you for someone who still understands my words without getting upset

At the layout I posted this resistor was meant to be the last at the lorlin step located in the frontpanel pcb which is not on the picture I posted. I managed to put a 5 ohm resistor there in my two channel version I have here for even more gain without obvious noise on both channels at the last step. But this seems to be too much for the four channel version because then it makes this behaviour at the last step.

I reduced these resistors at every channel to the recommended 10 ohms again and the channels are stable now so there was no need to throw everything of that layout in the trash or in doubt. Too be honest the gain is totally enough for every real world usage. If you need 60db of Gain for your lovely ribbon mic or whatever it will be not the best recording in the world anyway imo. As I don´t want to win the next tec award with this little micpre project I can live with this so far and can take a look in the mirror without feeling ashamed

Some posted somewhere that their design could achieve even over 70db of gain so I thought this would work. Maybe this was my mistake to begin with and it did become a problem when I combined two 2 channel boards.
As I wrote I will try a better layout with all four pres on one board with a bigger gnd plane and additional caps to see if it performs better.

Again I only posted this old channel layout of that 4pre because I built it years ago and remembered it also has this problem even with the 10ohm resistor placed in, so I think my new board has a much better grounding because they obvious perform better.

I throwed this old 4prepcb in trash as I remember, and made a new board on my own with two channels and bigger grounding areas. I only build these 2 channels then in the last time so I never had that problem. It was only now when I combined these to get 4 channels in one 1HE unit that they show that and that was the same prob I had with the 4pre design I posted. I hope it is clear now once and for all.

So it seems to be a combi of tracing layout, grounding and gain limiting which could produce this phenomena which is totally logical.

 

[Bo Deadly]

"The gain network is pretty low Z at high gain so I don't think it's an immediate source of instability."

Oh really?

https://thatcorp.com/datashts/Analog_Secrets_Your_Mother_Never_Told_You.pdf pdf pages 16-18.

And I'll ask again: "Maybe I missed it on the layout but where is R8 the 10Ω gain limit resistor in series with the 1000 µF?
If missing you have no maximum gain limit stop resistor."

Ok. I was wr-wr-wrong.

Even though the traces are not going to account for an imbalanced of several pF, clearly the gain control net is highly sensitive to common mode imbalance. So that output trace running next to one leg of the gain control net is very possibly a source of instability.

And more generally this is something very important to understand when dressing wires to a gain control pot for example. If it's longer than a few cm, it should be a fully shielded two conductor. Or better, just integrate the amp into the panel right next to the pot and skip extraneous wiring.

 

[user 133392]

If it were me (next time) I would take that 10Ω on the front panel board and make it two 5Ω resistors each located close to the 151X's pins. That will help isolate capacitance on the gain pot side of the traces/leads and improve common mode rejection for both the input and what's picked up on the Rgain lines. Having the 10Ω on the board is a good second option but splitting it is better.

Joe Neil, at my recommendation, put inductors in the Rgain lines to provide a ultrasonic/RF low-pass filter in his preamp: Wayne's Mic Pre photos (By Joe Neil aka Bruno2000) - Pro Audio Design Forum (I would do the servo in this differently now.)

The Rgain nodes are perhaps the most sensitive node in the entire preamp.

Electrical balance and symmetry along with physical balance and symmetry is your friend.

"And more generally this is something very important to understand when dressing wires to a gain control pot for example. If it's longer than a few cm, it should be a fully shielded two conductor. Or better, just integrate the amp into the panel right next to the pot and skip extraneous wiring."

Yep. I like the second option best.

The first option would work better if the shield was driven by the common mode signal.
For the 151X that has to be derived from Rg+ and Rg- or the input since we can't get it from the output.
For an op amp or discrete preamp where the outputs are available prior to the CMR stage the best option is two individual STP with the inners connected to Rg and the shields driven by the outputs.
The cable capacitance then appears as a parasitic feedback capacitor which provides benefit.

 

[JMan]

No technical advice here, but since you’ve brought it up more than a few times now and it has been politely ignored — everyone here knows Gustav and his pcbgrinder and Goly endeavors. He’s a member of the forum. He’s wonderful and makes great equipment. With that said, there do exist people out there with a greater wealth of knowledge and insight accumulated through decades more experience in the field, and I dare say you are (perhaps unwittingly) picking fights with more than one of them here as they try to offer help. It’s great that you can reference Gustav’s work, but it’s possible you may be placing that project’s virtue on too high a pedestal in insisting that your issues could not stem from something that you see in common with it.

 

[JMan]

Separately, I saw your comment in which you mentioned that you are working up a new pcb layout with a ground plane. I am far (like, really far) from an expert on this, so take my suggestions with the appropriate grain of salt, but I see two minor things that might be easy adjustments that could be helpful.

First, with the routing of your traces (in addition to the notes about capacitative coupling), I have been advised in my previous efforts to design pcbs to try to cross traces at a 90-degree angle, whereas I see numerous 45-degree crosses in your layout. This may or may not be useful advice, but it does seem like a very easy change.

Second, and this isn't related to noise, it looks like you've got relays on the board but no flyback diodes. Here again, the advice I've internalized is that it's good practice to include them. You appear to have the space, so perhaps it's worth considering just as a protective measure for your relays?

 

[Matt Syson]

Indeed the relays should have diodes to suppress the emf when switched off AND the control lines should be kept well away from audio and be from either a separate supply or resistor + capacitor 'isolated' from the audio rails. The coils of relays can couple magnetically into nearby audio tracks/components. Using a small choke as part of the gain set resistor chain can help reduce risk of oscillation as it would roll off hf gain slightly.

 

[user 133392]

While we're changing things I would suggest you reconsider the series capacitor in the THAT 1646 input.

I realize the intent is a switchable HP filter but there are two problems doing it that way:

The first is the input impedance of the 1646 can vary by as much as +/-20% which will produce lot-to-lot variations in corner frequency.

Secondly, when the HP filter is active, the reactance of the capacitor will upset the low frequency balance of the output. The 1646 input has to be driven by a low impedance source. You can put a coupling cap there but its reactance needs to be small and the value large.

A simple way to add a first-order HP filter is use a servo from the 1510 output to the reference pin as shown in the THAT application notes. If you shift the servo LP frequency higher into the audio band, using a switched resistor, the derived HP response becomes your filter. You don't really need the servo for DC removal since you already use a Cgain but having the servo provides an opportunity to eliminate it. Or just use the added op amp for a real HP filter.

When you do PC layout using a THAT 1646, be sure to keep the ground connection, pin 3, super-short, beefy and above all non-inductive. There's not much current flow out of it, but inductance in series with pin 3 will make it oscillate.

 

[timpanic]

Joe Neil, at my recommendation, put inductors in the Rgain lines to provide a ultrasonic/RF low-pass filter in his preamp: Wayne's Mic Pre photos (By Joe Neil aka Bruno2000) - Pro Audio Design Forum (I would do the servo in this differently now.)

That was really that kind of feedback I was looking for. Can be done when every module is decoupled and grounded right and the gain limits are respected.

Thx for that post. You can clearly see on the pictures that at this micpre the modules are indeed connected in series at the back of every modul with normal wire thickness and wago connectors as I mentioned at the beginning of this thread to put this out of the equation and got beaten up for it badly. Suddenly Gustav and I were somebody who have no clue about magnetfield and electronics in generell. And then the people wonder about the tone here or how it comes across. I am still under the assumption that some people have a will to misunderstood things too quick to produce themself. If that is not the case I am sorry but if I must hear their view here was mine.




I am just old enough to know that "you have to" knowledge alone is not helpful when other things are missing. Like the biggest studios who think they must have this and that and you MUST do it this way etc..and then when the press play everything sounds like crap and the demo recorded with a mackie has more soul. Surely soul doesn´t help with noise but you get the point.

I once visited a big famous top ten no 1 hit charts proven studio in Hamburg which was for sale. An old hired technician from that place was there who knew everything about every op amp replaced and modification they did to all their PSUs, gear, monitoring, studio star groudning, isolation, electricity, wires used over the decades etc ...end of the story: everything was defekt. The console was humming, the cabling a mess. Some main grounding was missing completly at some racks because somebody thought this is better. The machine room looked like after an explosion. So, I learned a lot on that day about theory and praxis.



Also as someone here mentioned my tone became ironic because I just wanted to defend an old design ( even faulty with lousy grouding maybe) from gustav, which was my intention in the first place. So Gustav, if you read this, I am sorry letting your old design getting feed to the Lions I did not know I can tell you I once had an ALPHA CHANNEL from SSL on my bench which was much noisier at zero gain than what I have here now and that was a commerical product.

As I said, I have solved "my" problem on my own with my own pcbs so far but I guess I would better never post any of my designs here after seeing how fast people declare people as idiots who have done so much for the forum and the whole diy community.

 

[timpanic]

Yet again Timpanic has failed to understand. Using the expressions 'fine and quiet' are a million miles from 'I have tested my design using industry recognised procedures and can now confirm that my implementation, using the specified chips (whatever they may be) conforms to the standard that is expected by the chip's manufacturer. failure to read and understand all available documenbtation about the components (chips in this case) and then comply with the various constraints. The 50 Picofarad 'limit' on the output of many chips is a common mistake as is many attempts to make 'your' piece of gear super wideband. DC to MegaHz for a mic preamp is simply asking for trouble unless you have made very careful the whole design can manage to do this and would be happy with modulated RF being applied via the mic input connector. As mobile phones work in the Microwave frequency bands, any conductor of around 6 inches or so becomes an aerial so many aspects of 'your' design have to allow for this possible intrusion. Of course you can't hear (and would probably struggle to measure) the RF carrier of mobiles for example but depending on the modulation you CAN receive assorted chirps which will be amplified/demodulated. Many years ago I had to 'RF proof' a mixing desk against having a 2 watt 'walkie Talkie' laid on it's control surface (on top of the knobs). It required relaying all the circuit boards that others had designed. star point 'grounding' is a fantastic concept but difficult to achieve in the real world where everything has a physical size so your single 'point' has to become a network of interconnected points.

Speechpolice as hell. What an ugly comment.
You will never know what I understand. Ah Suddenly I have a DC to MegaHz problem. Okay. Omg. Better give up about anything with electronics and crawl back in the cave.
FINE AND QUIET COMPARED TO OTHER NON DEFEKT COMMERICAL BUDGET MICPRES. And now? Helicopters? Squad Troops?
I use the term I want and I thought we are here at "GROUPDIY" when I look at the top of my browser and not and the last semester of the electronic university. If I had "tested my design using industry recognised procedures and can now confirm that my implementation, using the specified chips (whatever they may be) conforms to the standard that is expected by the chip's manufacturer..:" I would have never posted any question at all but this is too high for you I guess, because you´re only here to show what a hell of dude you are.

Next time I will only post numbers without any words. With this attitude nobody ever would have built any stompbox in the world.
What an arrogant comment. I am out and don´t want to waste my time with this anymore here. I would have deleted this whole thread but I can not. So admin please, throw me out of the forum and delete this thread.

 

[Matt Syson]

Oh dear!

 

[user 133392]

They call PC board layouts "artwork" because it is. (And sometimes isn't.)

 

[MidnightArrakis]

Speechpolice as hell. What an ugly comment.
You will never know what I understand. Ah Suddenly I have a DC to MegaHz problem. Okay. Omg. Better give up about anything with electronics and crawl back in the cave.
FINE AND QUIET COMPARED TO OTHER NON DEFEKT COMMERICAL BUDGET MICPRES. And now? Helicopters? Squad Troops?
I use the term I want and I thought we are here at "GROUPDIY" when I look at the top of my browser and not and the last semester of the electronic university. If I had "tested my design using industry recognised procedures and can now confirm that my implementation, using the specified chips (whatever they may be) conforms to the standard that is expected by the chip's manufacturer..:" I would have never posted any question at all but this is too high for you I guess, because you´re only here to show what a hell of dude you are.

Next time I will only post numbers without any words. With this attitude nobody ever would have built any stompbox in the world.
What an arrogant comment. I am out and don´t want to waste my time with this anymore here. I would have deleted this whole thread but I can not. So admin please, throw me out of the forum and delete this thread.

[Next time I will only post numbers without any words] -- Go onto YouTube and look-up "Numbers Stations"!!! Maybe that's where you should actually belong!!! NOTE: "Numbers Stations" are really weird, strange and "unknown" radio stations throughout the world!!! Who knows what it is that they are doing!!!

/

 

[Newmarket]

Speechpolice as hell. What an ugly comment.
You will never know what I understand. Ah Suddenly I have a DC to MegaHz problem. Okay. Omg. Better give up about anything with electronics and crawl back in the cave.
FINE AND QUIET COMPARED TO OTHER NON DEFEKT COMMERICAL BUDGET MICPRES. And now? Helicopters? Squad Troops?
I use the term I want and I thought we are here at "GROUPDIY" when I look at the top of my browser and not and the last semester of the electronic university. If I had "tested my design using industry recognised procedures and can now confirm that my implementation, using the specified chips (whatever they may be) conforms to the standard that is expected by the chip's manufacturer..:" I would have never posted any question at all but this is too high for you I guess, because you´re only here to show what a hell of dude you are.

Next time I will only post numbers without any words. With this attitude nobody ever would have built any stompbox in the world.
What an arrogant comment. I am out and don´t want to waste my time with this anymore here. I would have deleted this whole thread but I can not. So admin please, throw me out of the forum and delete this thread.

Are you okay ?

 

[JMan]

Well this has been a rather fraught thread. Timpanic, we all know that internet conversations can be very difficult to interpret in terms of tone, and I think everyone has sensed that you are coming in pretty hot and brash, while you have clearly felt that much of the response that you received was unhelpful or even hostile. As Ian mentioned earlier, it's tough (even for great minds like his) to diagnose a problem remotely, and all the more so when one doesn't know where the person on the other end is coming from. For example, I doubt I have anywhere near as much experience as you do in electronics, but I've been on the forum for a couple years and had enough opportunities to prove my ignorance, so there's a better chance that people might know how to meet me where I am. Being that you're new, your claims of experience are as-of-yet unproven (which doesn't mean they're untrue), so folks have been taking guesses at what you might or might not know. It seems like you took offense to this ("Suddenly Gustav and I were somebody who have no clue about magnetfield and electronics in generell"). And again, mentioning someone who is known like Gustav is fine, but you are not Gustav -- folks don't know who you are yet. I hope you'll stick around long enough to let us get to know the side of you that isn't represented in this thread, and for you to see that what clearly hasn't been the forum experience you wanted isn't "just how it is here." Anyway, I'm glad you solved your issue and I wish you the best of luck.

 

[JMan]

Also, MidnightArrakis is right, number stations are fascinating (and more than a little creepy)!