Double balanced microphone preamp project

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

chrissugar

Well-known member
Joined
Jun 4, 2004
Messages
1,315
Location
EU
As most of you know, I'm talking about the topology of Graeme Cohen that was published in AES 2106. I don't remember if the document was ever posted so here is a link from where you can download it:
http://www.seventhcirclestudios.com/SCA/C84/docs/2202.pdf
For a while I started gathering information about this topology and it's various implementations like the Millenni@ transformerless pre, Fred's jfet version:
http://www.forsselltech.com/JFETMP1.PDF
and Tim Ryan's version:
http://www.seventhcirclestudios.com/SCA/C84/docs/c84_sch.pdf

Interesting is that for a long time I had the same idea as Tim to add a Jung buffer that can drive anything. Although my idea is to put it outside the feedback loop like I saw in many audiphile preamps where the output buffer is after the global feedback.

I have some questions. In the original paper and in Tim's circuit there is no capacitor in the gain set circuit, but in Fred's version there is a 3900microfarad capacitor. My question is: is this capacitor necessary or not. It has a similar role like the ones used in the gain circuit of instrumentation amplifier type mic pres? It works like a high pass filter to remove DC or it is present only to avoid some DC on the gain pot?

My next question is in the Millenni@ manual ( http://www.mil-media.com/pdf/hv3c-manual.pdf ) they say the headroom is 32dBU but at the PSU voltages inside I have no idea how is that possible.
At +/-24V the maximum output peak-peak voltage is lower than 48V in ballanced mode, and that would be not more than 26dBU. Anyone has an idea? It is a mistake in the specs?

Another aspect is what kind of EMI/RFI protection would be suitable for this project? I'm more inclined to build a variation on Fred's version because of the Jfet inputs that are less problematic with EMI/RFI. Any thoughts?

chrissugar
 
quote: "I have some questions. In the original paper and in Tim's circuit there is no capacitor in the gain set circuit, but in Fred's version there is a 3900microfarad capacitor. My question is: is this capacitor necessary or not. It has a similar role like the ones used in the gain circuit of instrumentation amplifier type mic pres? It works like a high pass filter to remove DC or it is present only to avoid some DC on the gain pot? "


Both. If you are using a pot it's essential. If you can tolerate d.c. offset maybe it's not, but that d.c. offset may be substantial. Cohen got away with gain down to d.c. by using the LM194, but virtually anything else will have too much offset and drift unless you block d.c. somewhere, or equivalently sense and servo out the offsets.

Note also that you will need two buffers, one after each output diff amp.

Thanks btw for the link. I had not read this paper before.
 
Thanks btw for the link.
And I just payed five bucks for it yesterday. :evil:

I don't see the problem with the 32 dBu out; assuming we have OPA604s running @ +/- 24 V, that makes about 22 Vp out, with a balanced out makes 48 Vp, that's almost exactly 32 dBu.

Did you check the MicPre-Meta for RFI-protection-threads (can't help advertise it a bit :wink: )?

Samuel
 
[quote author="chrissugar"]I don't remember if the document was ever posted so here is a link from where you can download it:
http://www.seventhcirclestudios.com/SCA/C84/docs/2202.pdf
chrissugar[/quote]

You have not seen it before because it is a violation of copyrights without the permission of AES. Tim should know better.
 
Thank you all for the answers.

[quote author="bcarso"]
Both. If you are using a pot it's essential. If you can tolerate d.c. offset maybe it's not, but that d.c. offset may be substantial. Cohen got away with gain down to d.c. by using the LM194, but virtually anything else will have too much offset and drift unless you block d.c. somewhere, or equivalently sense and servo out the offsets.

Note also that you will need two buffers, one after each output diff amp.
[/quote]

I supose if you use some 2SK389 in paralell in Fred's version, DC ofsett can be low. Also there are two trims at the output stage for DC trim.
Yes, I know that I will need two Jung buffers. :grin:

[quote author="Samuel Groner"]I don't see the problem with the 32 dBu out; assuming we have OPA604s running @ +/- 24 V, that makes about 22 Vp out, with a balanced out makes 48 Vp, that's almost exactly 32 dBu.[/quote]

Just take a look at this table: http://www.uneeda-audio.com/dbtbl.pdf
At 26dBu you have 43.73volts peak-peak, even at +/-24V you can't have at the differential out more than 48volt peak-peak. This is not like an API preamp that can produce a higher output voltage then the PSU voltage because of the output transformer. 32dBu would be almost 88v peak-peak. Please anyone can explain? :?

[quote author="Samuel Groner"]Did you check the MicPre-Meta for RFI-protection-threads (can't help advertise it a bit :wink: )?
[/quote]
I read the thread but I was curious if anyone can recomend something specific for this topology.

chris
 
quote: "I supose if you use some 2SK389in paralell in Fred's version, DC ofsett can be low. Also there are two trims at the output stage for DC trim. "

Can be low.... Who knows how good the Toshib* SK389's really are? If I am making a bunch of something I have different courses of action. One thing is to build a bunch and see what the statistics are on offset and temp drift---and pray that this sample tested is representative. Another way that may be safer and giving less opportunity for career limiting is to take the manufacturer's data and see how bad off I could be.

There's a third way which is to select a bunch of parts, test and qualify them, and hope I don't run out. Tedious and high-risk, significant investment of time and money as well. If we are making a few of something like the usual case here, maybe it's a reasonable approach.

If we do approach number two with the 389 we pull out the datasheet. It tells us that the differential gate-source voltage for equal drain currents in each half is EDIT 20mV. Ouch!---at the 50dB gain that will be an output offset of over EDIT 6 volts. But then you say we trim that out. So how much do things drift over the reasonable temperature range in a studio rack or wherever this is to be deployed? What's the drift spec? Ouch!---there is none!

We can turn to books to help us predict the drift based on the initial offset, and the tempco after trimming, although some of these arguments have a lot of assumptions that may not hold. We can hunt for other parts with guaranteed specs and hope we find something suitable (our friends at Linear Integrated S*stems might be consulted, since their [still anticipated?] 389 clone is of interdigitated construction, which may equate to lower offsets and drift).

Or we can simply put a big ugly 'lytic in the input circuit, or use a servo, or decide the d.c. can be tolerated, or...
 
Thank you Brad

Wow, 3Volts! :shock:
It looks like the Jfet version really needs that big capacitor.

chrissugar
 
Sorry, but with two posts in a row from Chris I just have to ask: have you seen the new W&G yet ? I did last weekend, and it was :thumb: :thumb:
 
[quote author="chrissugar"]Thank you Brad

Wow, 3Volts! :shock:
It looks like the Jfet version really needs that big capacitor.

chrissugar[/quote]

See EDITs! I didn't read the datasheet right---it's twice as bad!
 
[quote author="clintrubber"]Sorry, but with two posts in a row from Chris I just have to ask: have you seen the new W&G yet ? I did last weekend, and it was :thumb: :thumb:[/quote]

What's the "W&G"?
 
[quote author="chrissugar"]Brad,

Is my calculus for the headroom correct?

chrissugar[/quote]

Well let's see. 32 dBu is 10^(32/20) * 0.775V rms = 39.8 * 0.775V = 30.85V rms? Then we have to swing square root 2 times this, peak-to-peak, on each output, so that the net balanced output is twice that. Sounds like we need to swing 43.63V, or +/- 21.82V, on each output, just doable with 24V rails if the signal is zero d.c.
 
Oh, I see where my error was. :oops:
The less then 48v p-p would be for unbalanced out but with balanced you add 6dB.
Thank you.

chrissugar
 
> at +/-24V you can't have at the differential out more than 48volt peak-peak.

At one instant, pin A is up at +22V, pin B is down at -22V, differential voltage is 44V. If pin A is marked "+", then we call it +44V. At the next half-cycle the situation is reversed, -44V. Peak-to-peak is 88Vpp, or 31.1V RMS, or +32.07dB above 0.775V RMS, the accepted reference for zero dBu.

Would I call that "headroom"? Not until I define a reference signal level, which usually isn't zero dBu in studio custom.

I have doubts about 5532s living-long on +/-24V: that's over spec, and the 5532 spec is on the high end for the process it must be made on. I have doubts about 2V loss in the 5532 output stage: it is mighty clever and sure can swing far with light loading, but in real life it may be over 2V loss. +/-21V rails with 3V loss gives "only" 25.4V RMS, a whole 1.7dB less than +/-24V rails with 2V loss.

Would I worry about "headroom" from a 31V (or even 25V) RMS output? No, I'd worry about smoking inputs; not in normal operation, but in a screw-up.
 
I read the thread but I was curious if anyone can recomend something specific for this topology.
I'd do it similar to the C84; looking from the mic side, first a 1 nF cap across hot and cold; then a common-mode chocke (say 39 or 47 mH), bypassed with 6k8 resistors, then two matched 1 nF caps to chassis.

BTW, I believe that the C84 is darn close to what we get from M*llennia.

Samuel
 
[quote author="bcarso"][quote author="clintrubber"]Sorry, but with two posts in a row from Chris I just have to ask: have you seen the new W&G yet ? I did last weekend, and it was :thumb: :thumb:[/quote]

What's the "W&G"?[/quote]

Sorry, a bit cryptic.

W&G, as in Chris his avatar:
http://www.wallaceandgromit.co.uk/fla/wg.html

http://www.wandg.com/

newhome_01.jpg
 
Wallace & Gromit is a rush!! I have been following them for years. However the irony is that the very day that their movie opens in the States as the top movie, the warehouse storing their sets and memorabilia burns to the ground! Argh ...
 

Latest posts

Back
Top