[DESIGN] Original(?) mike-amp design

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Will this evolve into a fet design? bjt's are more bullet prrof and easy to get and cheap.
Can't wait for the thermal analysis!
Everybody should create a PRR directory if you hane not done so already, and cut and paste this stuff!
Thanks PRR!
cjj
 
[quote author="cjenrick"]Will this evolve into a fet design?[/quote]I'm betting that it will stay BJT. PRR just seems like that kinda guy...we'll see!
Everybody should create a PRR directory if you have not done so already
WAAAAAAY ahead of ya CJ! But that's good advice to all!

Looking forward to the next lesson!
Charlie
 
> your circuit looks something like some of the PASS amps at passlabs DIY section.

We all stand on the backs of giants, or at least step in their old ideas. Nelson does it too, but he has vision and timing and the guts to explore relatively new territory. There is nothing in my plan that wasn't "state of the art" in 1972, except "the art" was heading in other directions then.

> You're building a discrete SSM2017 (the input stage)

There are not that many ways to build a good (low noise, good CMRR) transformerless mike input. You start with two BJTs, Bases to the input, Emitters to the feedback, and Collectors running off through some gain.

The output stage on the usual implementations is important. My plan has a glitch because I have omitted it. It is probably a workable glitch, but the kind of thing that would confound most buyers. That's another reason this is probably stuck in DIY-land. OTOH, omitting the output stage AND the added stage needed for balanced output does shorten the signal path.

> Hope I am not putting the cart before the horse here!

If that horse ever limps in, you will see that:

> a "gain block" which could made up on small boards and then maybe potted

Maybe in a lobster-pot. Or "pot it an old muppets lunch box". I'm headed to a VERY extravagant design, LOTS of heat.

And quite specialized FOR the low-Z mike input and Pro-output situation. Not general-purpose utility amp. A one-trick pony.

> would the usual input add ons like pads, rf filters, phantom supply etc have any effect on the noise level

Yes.

But one neat trick is that pads will rarely be needed, since it goes to unity-gain and can do low-gain with high S/N.

And phantom should not have a significant effect on noise (though I have seen that said). 13,600 ohms across 200 ohms is "nothing".

RF is a problem and I admit my (to be described) values are unproven and may not be suitable in high-RF areas. And if you just try to flesh-out the above sketch, it is likely to be an RF transmitter (in the 100MHz range) rather than an audio amp.

> all TO3 packages...

I had TO220 outputs in mind. But you can use TO3.

> the biggest op amp ever.

Top of my head, there are a dozen production op-amps bigger (more power) than this.

And this isn't the classic general-purpose op-amp. It is a differential amp, with the feedback (operation) tightly woven into the design.

> Will this evolve into a fet design?

No. To my mind, FETs are not optimal for low-Z mike inputs. Some wise designers disagree, which means they know more about FETs than I do. There are many ways to do it. This is mine.

> Can't wait for the thermal analysis!

Actually, laying-out the resistance paths is harder. There are paths where 0.25 ohms stray resistance will throw the gain way off, and similar small inductance will make it puke radio waves. And these nodes are on the ends of big hot resistors, not far from transistors that should not be hot.
 
Actually, laying-out the resistance paths is harder. There are paths where 0.25 ohms stray resistance will throw the gain way off, and similar small inductance will make it puke radio waves. And these nodes are on the ends of big hot resistors, not far from transistors that should not be hot.

This is exactly the problem That Corp is having according to there engineer that was at the Berlin AES. They still did not have the right layout in there 1510 chip to get the specs they wanted yet.

All modern mike preamps have to have a +48V supply for Phantom. It seems silly to also have various other voltages like +/-15V. You can build a fine preamp with just +48V. The preamp will take more power than just the Phantom needs, but it is easier to beef-up the one supply than to build several supplies. Single-supply tends to force coupling capacitors, but Phantom already forces the use of caps. (Or transformers, but see above; or very clever common-mode design.)

Our TMP8 8 pack of Mic Pre runs on one 48v rail. It makes sense since the only part of a mic pre that has a standard specification with respect to volts is phantom power.

Joe
 
PRR a high current large area device gain stage sounds like a very very cool design!

I posted about the pass stuff so people could read up some of the concept. I find I tend to like BJTs better than fets for circuits that can use BJTs They are easier to "get right" because circuits can often be designed so they are not as picky as a fet circuit. Plus large area mosfet have a weird gate charge step responce and they often need a high current drive to sound OK IMO.
 
Wonderful !

Thanks PRR for the explanation/insight of the circuit.

(as he reads it over and over again :oops: )

A seperate forum is ok with me or if we just keep it here is ok too!

Z
 
[quote author="JLM Audio"]This is exactly the problem That Corp is having according to there engineer that was at the Berlin AES. They still did not have the right layout in there 1510 chip to get the specs they wanted yet.
[/quote]

yep
... and will we ever see the 1510 chip a reality

and YES bluebird,
I think it is time the professor came back to the class ... :wink:
the children are getting restless
 
I stopped at THAT's booth at AES last year and asked about "Bill Whitlock's circuit" since I didn't know the part number offhand. (It's the 1200 series). The guy groaned audibly. I must have been the 100th person who'd asked about it that day...

Of course, I realize that you guys are talking about the 1500 series.

http://www.thatcorp.com/relsched.html
 
> BJTs... are easier to "get right" because circuits can often be designed so they are not as picky as a fet circuit.

It is also easier to get in trouble with high-gain BJT instead of mild FETs.

I finally realized what some of my simulation results were telling me: my plan has a noise problem. Not at high gain, but pretty bad at low gain, which was the original idea!

You can't trick Mother Nature.

I'll try to find a fix, but everything I've looked at either won't work, is very expensive, or both.
 
If you manage to muscle through this, you might even teach your self something. :grin:

when you say expensive do youy mean transformers? This might not be the worst thing. I know the point was to make something cheap and simple, but just simple might do for expeimental builders like me... who will build something just for the joy of soldering stuff together.

The circuit may not be living up to the "original idea" but maybe the idea could change?
 
Hi PRR

Any news about this project? I'm very curious about it. Did you do any tests?

chrissugar
 
How about running four or eight transistors in parallel on the input per side to lower input noise?
Low noise, high gain BJTs are really cheap these days. You can get one hundred MPSA18 or BC550C parts for less than $10.
I am thinking that when you use eight in parallel matching individual transistors for both sides becomes less important. Mix them up in a big bag and through random distribution they should end up pretty close. :?:
 
> How about running four or eight transistors in parallel on the input per side to lower input noise?

That's a thought. Want to make something of it?

I will point out that a couple switch-transistors per side, or one LM394, can fairly easily get down below 2dB noise figure. Further improvement is about inaudible.

And the LM394 ($6 at DigiKey) will match sides better than any shaken bag of same-lot devices.

And this does not solve the problem I was attacking. The feedback resistors add noise. At high gain, you can use a very small shunt resistor, under 100Ω, and be insignificant. But we may use the same preamp with a big-head mike up-close on too-loud percussion, so we need very low gain settings too. With conventional feedback resistor values, noise rises when gain goes below about 40dB, and can be quite large at 20dB. It is possible for mike-amp noise to be as large or larger than room noise, mike noise, or recorder noise. And it shouldn't have to be that way, though I admit I'm not happy with where I thought I was going.
 
I'm not sure I understand this.

In the feedback path, you have a fixed series resistor, and a variable shunt resistor. For low input signals / high gain the shunt is small, so the circuit is optimised for noise.

For high input signals / low gain, the shunt is larger, so that higher resistance (in extreme, just the series resistor when we're at unity gain) will contribute more noise at the input stage. Is that what you mean?

But in that second case, the _gain_ is also much lower, so the noise that comes from the feedback resistor to the input is also less amplified. So I can't see why the noise level (at the output) should be increased.

"Noise figure" may be worse, but that's just a number. The noise at the output should be lower, the SNR increased, with the higher input and lower gain.

I can see that there is still room for improovement (i.e. making the SNR even better in the high input / low gain case), but I don't understand the worry about lower gain increasing the noise, despite the higher resistance.

Am I missing something?

JH.
 
Jurgen,

You're right that the absolute noise level will not be higher at low gains, but at low gains you would prefer to have the benefit from even lower noise as well..

(edit..)

Jakob E.
 
[quote author="gyraf"]Jurgen,
benefit from even lower noise as well..
[/quote]

Ok - then it's clear.

JH.
 
ok, I've re-read this thread a few times and am finally understanding what I am seeing in PRR's design, however, I'm completely ignorant of certain things pertaining to Preamp design...

Am I correct to assume that by paralleling BJTs that you are sharing the currents flowing through them which are causing the noise? higher current=more noise?

do lower noise BJTs mean that they are also less resistive(or vice-versa)?

what characteristics are you looking for to determine if a BJT is suitable, other than just sticking it in there and trying it/simulation? Is there something specific on a datasheet to look for? Is voltage rating any kind of factor in current based noise?

ok, on to my last question.... How does one figure for feedback? I assume it's based on what you design, but any more insight than that? i also assume there is a formula for feedback resistors?

well thanks for any answers you can give!

:guinness:
 
One of the thing I have been meaning to do is to try devices like the tip29,30,31,32. They are power devices with a larger die.

I have built a tip29 base Si rangemaster guitar boost petal low noise for a treble boost guitar petal. I wonder how a matched pair running at a few Ma would sound as a front end for the circuit in this thread and the green etc type circuits?
 

Latest posts

Back
Top