Help with noisy preamp - replacement parts?

Hi, long time lurker on this forum, first time poster.

Today I acquired two Australian Monitor AMISPre-1 Mic amplifiers.
I opened them up and see they use one JRC4558 Op - Amp and two A733 PNP Amplifier Transistors, these are common parts and as it is a quite simple circuit I think I should be able to upgrade some components in the signal path to lower-noise components.

I have a couple of OPA2134s and 2N3906s and I am thinking of soldering in some sockets and trying out a few different combinations of opamp/transistor to see if I can lower the noise floor a bit.
Do you think its the active components or the capacitors in the signal path I should be focusing on the help minimize of noise?

It also has a 12v regulator in it but says the it should be run from a 12v supply?
From my understanding of voltage regulators you need a couple of extra volta above the output voltage to create a good output. Possibly that could be a big source of noise?
Any advice would be greatly appreciated!


Schematic link : https://drive.google.com/open?id=1K-x_fPtGbJwK2t4yzLULtuIP9wMN6sfN


Cheers
Liam

With the right transistors you might see a little bit of an improvement. Probably less so from changing the op amp because I think most of the gain is in the transistor bits. But you have to match them or you could get a significant offset on the output of the op amp. And there are transistors that have better noise characteristics than 2N3906. If you're going to find two matched, then you really probably need to buy 20 low noise high gain PNP and pick through them.

You might check for feed through from that voltage multiplier too.

Aren't the 4558 kinda hissy??? I don't have a problem with them.....Ashly uses them all over the place in their older stuff....

The schematic shows the DD version so maybe these are lower noise for this???....

@squarewave - thanks for the input!

Yeah It looks like the dual opamp is just buffering to unbalanced on one side and the other is doing something with the 12v to create phantom power. Did you mean if I change the opamp I might get crosstalk/leakage of DC to the signal side?

I soldered a dip socket in there and tried OP2A134 but no luck, must have different offset requirements or something because it clipped really easily and lost a lot of gain. Also beginning to suspect that the noise is coming from ageing A733 amplifier transistors. Next step is to socket them and see if I can get some drop in replacements for them which are lower noise.

@scott - It says JRC4558DD on the schem but on the board the chip is JRC4558A, could this be their sneaky way of cutting costs? I'm not too boned up on chip suffixes so didn't notice that. Maybe I could get some NOS JRC4558DDs and chuck them in and see if that helps before I recap/change transistors.

burstmembrane said:

@scott - It says JRC4558DD on the schem but on the board the chip is JRC4558A, could this be their sneaky way of cutting costs? I'm not too boned up on chip suffixes so didn't notice that. Maybe I could get some NOS JRC4558DDs and chuck them in and see if that helps before I recap/change transistors.

Click to expand...

. Since you have a socket now, maybe try some other chips????TL072 ??? I've used 4560 in place of the 4558 and they do sound cleaner to me......

If you have some odd, unreasonable noise, I'd be looking elsewhere as the issue  as squarewave brought up.....

What is happening exactly???

You're peobably better off building a decent preamp instead of polishing a turd.

You will get the most benefit from using premium low noise transistors in place of 733... (but I do not know how good or bad 733 is. I never heard of it before.)

The  733  are graded for beta (current gain) so highest current gain means lowest base current noise, so use the A733"K" (K means 300-600x beta).

This is a pretty common (well copied) topology.  So good low noise transistors and decent op amps will deliver respectable performance.  (note 3906 is not a low noise transistor).

JR

trashcanman said:

You're peobably better off building a decent preamp instead of polishing a turd.

Click to expand...

Seems an unnecessarily negative comment. The circuit is essentially good and noise performance can be improved by component type and value.

burstmembrane said:

@squarewave - thanks for the input!

I soldered a dip socket in there and tried OP2A134 but no luck, must have different offset requirements or something because it clipped really easily and lost a lot of gain. Also beginning to suspect that the noise is coming from ageing A733 amplifier transistors. Next step is to socket them and see if I can get some drop in replacements for them which are lower noise.

Click to expand...

The OPA2134 problem sounds wrong.
Is a TL072 okay in there ? Check the decoupling and voltage rails.

burstmembrane said:

Yeah It looks like the dual opamp is just buffering to unbalanced on one side and the other is doing something with the 12v to create phantom power. Did you mean if I change the opamp I might get crosstalk/leakage of DC to the signal side?

Click to expand...

Yes. It is slightly dubious that you're using one half of the OA for the mic signal and the other half for a high frequency voltage multiplier. That might bleed into the mic signal but probably not as hum. It would be more high frequency hash or maybe a whistling noise. You really need to characterize "noisy". Is it hiss, hum or something else? If it's hum, it could be a lot of things like grounding issues (high gain amplifiers are very sensitive to ground wiring), a ground loop or ground currents between the device and what it is connected to. Meaning it's very possible that the device is perfectly silent until you plug it into something (which I suppose is not very useful!). Hum would also be indicative of a power supply issue (where are you getting your 12v?). The larger electrolytic capacitors dry out after a few decades and loose there filtering capability. You might just need to recap. If the noise is hiss, the noise performance either just stinks or the signal you're putting into is noisy. There are so many things to consider, it really cannot be explored entirely in a few Internet posts.

burstmembrane said:

I soldered a dip socket in there and tried OP2A134 but no luck, must have different offset requirements or something because it clipped really easily and lost a lot of gain. Also beginning to suspect that the noise is coming from ageing A733 amplifier transistors. Next step is to socket them and see if I can get some drop in replacements for them which are lower noise.

Click to expand...

OPA2132 is actually probably not the right OA for this because for one thing it is a very fast op amp and that can lead to stability problems. You're trying to use one half in a circuit with probably 50dB of gain and then the other half to run a high frequency voltage multiplier. The RC4558 is much slower and that might be a factor in that oscillator circuit. If you really want to change the OA (again, I don't think it will really make that much of a difference because all of the gain is in the transistors) you could try a sturdy bipolar OA like NE5532.

Regarding the transistors, transistors do not age really. They can die. And they are born with certain characteristics like gain and noise performance. And in this case they MUST be matched because the input is differential (search "differential pnp circuit" in Google images).

Thanks for the responses everyone!

Sorry if I wasn't clear about the nature of the noise. The noise is preamp noise, as in white/pink noise occurring across the entire spectrum. There doesn't appear to be any oscillation in the circuit though at very high gain levels I feel like I can hear some sort of ringing which might be the transistors approaching oscillation? That's with gain pot at max though.
To be fair the noise floor is not insanely high. I'm just looking for ways to upgrade/experiment with the circuit for better noise performance.

There is pretty much no hum as the unit seems to be quite sturdily grounded/shielded with a chassis ground lug and a steel construction.  Have tried with a bench power supply (varying 12-15v), guitar effects wall-wart, 12v wall wart from Jaycar and a 9v battery and the hiss remains stable over these different power supply configurations which leads me to believe it is the transistors self noise which is being amplified  (not sure on the correct terminology) as has been mentioned.  I think the power supply is pretty well regulated/bypassed from what I can tell with the schematic.

I'll have to order some low-noise high gain transistors and give it a try. There seems to be a lot of types, do I need to match the specs of the a733 really closely in looking for a new part? The datasheet for A733 is https://www.mouser.com/ds/2/149/2SA733-310191.pdf.

Wondering if I'd be able to place a matched pair of low noise-high gain PNP's in there with the current circuit topology.

So far I am looking at 2SV888, MAT03 or possibly a matched pair from CoolAudio or That Corporation for stability. Might have to make a little breakout board or something down the track but that's fine. Or a 2n5087 which has a different pinout but I could make a breakout pcb.

Does anyone have any recommendations on low noise PNP's that might work in this application? I understand without actually having the circuit in front of you it'd be hard to tell but maybe just a prod in the right direction. I'm wading in a sea chrome tabs and datasheets haha. I could possibly put some trimpots in to rebias the transistors

@scott - I'm going to pick up some tl072s and ne5534's in my next parts order and pop them in. I had another look and it is in fact a JRC4558DD. I'm beginning to suspect the opamp might not be the main noise culprit at this stage though so might leave them in while I experiment with transistors.

Yes, 2N5087 is good. But NE5534 is a single OA. The dual version is NE5532. If you do change the OA you may need to add a 100n ceramic cap across the power pins. The faster OAs need good bypassing so low-ESR ceramic caps are sometimes required (although it your case it's only one cap because you're using single-supply and not the usual bipolar supply).

Thanks for the heads up re NE5534/5532. Up to my ears in parts numbers! I'm going to have a go at hand matching some 2N5087's using this method when my parts order comes in and see how they sound.

https://www.cgs.synth.net/modules/simple_transistor_matching.html

from a quick google search  2Sv888 looks like a low noise candidate while the 0.3dB claim is a little dodgy. http://www.garrettaudio.com/DATA/2SV888_DATASHEET.pdf

2sa1085E  http://www.el-component.com/bipolar-transistors/2sa1085

I never used either, My favorite for this application 2sb737s went obsolete last century. 

JR

There is nothing wrong with the original plan. Have you tried it *at the same gain* against other preamps?

It is possible the input devices have been abused, not to death, but to a degraded noise condition. This is not a job for puny devices like 2N3906 which are low-hiss at 10K source but not at 200 Ohms.  2N4403 is more appropriate and often used at this impedance level.

The opamp is NOT your hiss. The BJT input has gobs of gain, overwhelming whatever comes after it. While I do not like a fancy '741 for hi-level +/-15V audio, here the supply is 12V and slew limiting is much less. And changing the dual opamp risks trouble with the too-clever +48V supply. Go any further and it would be MUCH better to scrap this and build your own preamp.

Unless the hiss is truly a problem, *at same gain*, I say enjoy it for what it is. (Use it in your truck: the power is perfect.)

For the record all preamps have noise while a good one can be respectably below room noise etc, it will never be dead silent unless it is actually dead.

Low noise transistors can be degraded by allowing them to zener from roughly 7V of reverse bias across the base-emitter junction. In some designs this can be caused by phantom caps discharging or even system turn on transients.

Better preamps will include a clamp diode connected backwards from base to emitter to prevent the junction from ever reverse zenering. I notice your schematic does not have protection diodes, but this does not mean your transistors are damaged, only that they could be.

JR

PS: Yes the 2n4403 is lower rbb than 2n3906 but not considered low noise by modern standards. Back in the 70's/80s I saw the 2n4403 written up in a low noise design text (Motchenbacher and Fitchen IIRC). I think he paralleled a few of them.

Not to offend anyone with the obvious, but is your assessment of the preamp noise made with a proper resistive source? Rather than leaving the input unconnected, connect a 150-200Ω resistor to the preamp input (between pin 2 and pin 3) before you assess its noise performance. This is important with low En input transistors, as high input current noise usually comes along with low voltage noise and high bias currents, and without a proper (low) input source resistance, the input current noise will be overstated into the preamp's much higher input impedance.

Hi all,

Thanks for the perspective!

Maybe I was misleading in my original post. I think my quest for lower noise with these preamps will perhaps end in incremental change or a placebo effect. I suppose my motivation is less for silent preamps and more for squeezing as much as I can out of this design and also learning a bit about preamp topology and noise sources for building my own later on. 

Side by side with my Scarlett 6i6 and even Mackie 1202 preamp stages there is a noticeably higher noise floor. I haven't done any official testing but took some recordings side by side at similar levels with a Rode Nt1-A and they definitely were higher noise.

I am wanting to do some ambient/ nature soundscape recordings with these preamps and some spaced condnesers. I originally got them to replace my Zoom H4n pres which were very underpowered for this task and am intending to power them from a battery box/ dual 9v and regulator out in the field so low noise is quite important. The build quality is amazing and they are quite light and portable so good for this. Definitely would work with the car with an adapter!

I could use them as is for recording louder sounds or cityscapes and I would be quite happy. I have carefully desoldered the Opamp and A733 transistors so I can always go back if my experiments take me down a fruitless path. With the original components desoldered and socketed there is no noticeable change in the noise floor/performance so I think I didn't fry them (thank god).

I've ordered a couple of SSM2220 http://www.analog.com/media/en/technical-documentation/data-sheets/SSM2220.pdf
and MAT03EH Matched pairs - http://www.analog.com/media/en/technical-documentation/data-sheets/MAT03.pdf
from an supplier here in AUS as well as 20 2N5087s to have a go at hand matching.
Hopefully the supermatched transistors will help me tackle the noise floor.

@John, would you advise I add a zener from base to emitter in the course of my modifications?

So far my todo list is:

- Replace A733's with matched pair of MAT03EH, SSM2220 or 2n5087s
- Add Zener Clamps across Base Emitter to prevent reverse Zener Noise.
- Add Low-ESR Ceramic Bypassing to Opamp

Possible in future -
Replace resistors in audio path with 1% metal film
Replace ceramic caps in audio path with  film caps?


@Monte McGuire
No I haven't! I will make up a dummy connector tonight and have a go. Thanks for the tip!

burstmembrane said:

Hi all,

Thanks for the perspective!

Maybe I was misleading in my original post. I think my quest for lower noise with these preamps will perhaps end in incremental change or a placebo effect. I suppose my motivation is less for silent preamps and more for squeezing as much as I can out of this design and also learning a bit about preamp topology and noise sources for building my own later on. 

Side by side with my Scarlett 6i6 and even Mackie 1202 preamp stages there is a noticeably higher noise floor. I haven't done any official testing but took some recordings side by side at similar levels with a Rode Nt1-A and they definitely were higher noise.

I am wanting to do some ambient/ nature soundscape recordings with these preamps and some spaced condnesers. I originally got them to replace my Zoom H4n pres which were very underpowered for this task and am intending to power them from a battery box/ dual 9v and regulator out in the field so low noise is quite important. The build quality is amazing and they are quite light and portable so good for this. Definitely would work with the car with an adapter!

I could use them as is for recording louder sounds or cityscapes and I would be quite happy. I have carefully desoldered the Opamp and A733 transistors so I can always go back if my experiments take me down a fruitless path. With the original components desoldered and socketed there is no noticeable change in the noise floor/performance so I think I didn't fry them (thank god).

I've ordered a couple of SSM2220 http://www.analog.com/media/en/technical-documentation/data-sheets/SSM2220.pdf
and MAT03EH Matched pairs - http://www.analog.com/media/en/technical-documentation/data-sheets/MAT03.pdf
from an supplier here in AUS as well as 20 2N5087s to have a go at hand matching.
Hopefully the supermatched transistors will help me tackle the noise floor.

Click to expand...

I have never matched bipolar transistors for that topology. Matching will not affect noise, except perhaps for gain pot noise, but that is why the schematic uses a capacitor in series.  Further the 5087 is not a desirable choice IMO.

@John, would you advise I add a zener from base to emitter in the course of my modifications?

Click to expand...

As often happens I guess I wasn't clear.  NO you only need to add basic small signal diodes (like 1n914, 1n4148, etc ). The Problem is caused by the transistor's base emitter junction turning into a zener at 6.8V reverse voltage. A simple diode clamp prevents the reverse voltage from rising above one diode drop.  In normal operation these clamp diodes are themselves reverse biased so have no effect on the circuit.

So far my todo list is:

- Replace A733's with matched pair of MAT03EH, SSM2220 or 2n5087s

Click to expand...

I've never messed with transistor arrays for mic preamps  (except perhaps for experimenting with the iconic LM394 back in the very old days). Look out for too much capacitance. Of your short list I only recognize the 2n5087 that was considered a low noise transistor back in the very old days, so may not even improve upon the 733 (assuming you get the higher beta 733).

A quick glance at a 5087 data sheet reports a 2dB NF ("noise figure", or noise increase compared to a perfect noiseless device) at 3k Ohm...BUT microphones are more like 150-200 ohm source impedance so a 3k noise figure spec is not in the same ballpark.

- Add Zener Clamps across Base Emitter to prevent reverse Zener Noise.

Click to expand...

small signal diode clamp.

- Add Low-ESR Ceramic Bypassing to Opamp

Possible in future -
Replace resistors in audio path with 1% metal film
Replace ceramic caps in audio path with  film caps?


@Monte McGuire
No I haven't! I will make up a dummy connector tonight and have a go. Thanks for the tip!

Click to expand...

If you have more than one channel perhaps keep one stock so you can compare before and after... you could even null them, but that won't tell you which one is noisier. Maybe useful for chasing distortion or response errors. I wouldn't expect very audible differences from you additional changes but you tell us, better yet make empirical measurements that are more reliable than listening tests.

JR 

JohnRoberts said:

I have never matched bipolar transistors for that topology.

Click to expand...

Really? But a mismatch is going to cause an offset error that could limit headroom pretty significantly no?

Another thing I just noticed about this circuit is the 7812 regulator after a diode is going to require more like 15DC supply and not 12V.