Matching Transistors for Valley People MicPre

[bcarso]

The (possibly good) reason for using more transistors is series noise reduction, that is, lower equivalent input voltage noise. But to get the benefits you generally have to run more current through the total of them. With the same current density per transistor the voltage noise will go down as the square root of the number of devices. But at some point the parallel noise, or current noise, flowing in the source impedance and the feedback network components, will counteract the improvement due to lower voltage noise. The inflection point is where the source resistance is equal to the equiv. input voltage noise divided by the input current noise at a given frequency, which is known as the optimal source resistance. At this value of source resistance the contribution from noise voltage and noise current are equal.

Also, at some point, the impedance of the gain-determining R's will start to limit the noise reduction---dominated by the thermal noise of the small one at high gains. And then as you reduce those you may start to tax the ability of the op amps to drive the feedback network (less of a problem at high gain, which is where you are more concerned about noise anyway).

As I mentioned a while back I think, your input impedance will hold up for a while because of feedback. If there were no feedback it would be another story.

 

[CJ]

Ahh yes, now I remember.
It's all in the app notes.
Maybe I should read the web page more often.
Click on the word doc if interesred:

http://www.vacuumbrain.com/docs/valley_people.html

 

[daArry]

great stuff cj :thumb:

 

[CJ]

Thanks!
And cheap too. No iron to buy, just a pwr supply.

 

[CJ]

Here is another version of the VP.
I am using this one with the input circuit shown on the first schematic I posted.

The output may feed a 1:1 transformer, or a unity gain diff input opamp (buffer).

There is no gain pot shown here. You can adapt the orig. gain scheme. A cap feeding the pot may help eliminate thumps when the gain is adjusted.

Resistor values may need to be adjusted depending on which transistors are used. The 360 ohm and 2.2 cap are kind of a mystery to me. Which side of the resistor should I run the neg. opamp pwr pin to?
Thanks!
cj

 

[daArry]

pfft

 

[tk@halmi]

CJ,

How is this pre going? I am curious if you had a chance to pass sound through it.

T

 

[CJ]

I will burn the Rev 3 board tonite. This project does not lend itself to breadboarding, so I am burning boards and stuffing them as I go along. I think I have something viable now. I am tweaking the current source, which has turned out to be a cross between the first schematic and the second. I found the diodes to be too thermally sensitive. This may just be due to the packaging, (glass, as opposed to transistor plastic), as both the transistor and diode are just p/n junctions on silicon. But the three transistor current source will be easier to couple thermally, as I can lay the transistors against each other. Interesting to note, if you heat up the two curent source transistors, the current provided to the array goes up. But when you heat up the diodes or trans diode, depending on which you use, the current goes down. So if the three of them are coupled thermally, they compensate each other to provide a steady current source despite tempeture changes.
It seems to me that the current source is more thermally sensitive than the actual array. Heating up the transistors in the array does not cause as big a shift in current.

And tweaking the bias resistor that goes from the bases of the current source to the neg rail really has a big effect on the circuit parameters, such as collecter voltage, etc. You have to set up the transistors you use with this resistor in order to get that -7.5 volts on the collector, and the slightly positive voltages on the emmiter and base.

I have the chassis all ready to go, so as soon as I get the board where I want it, it will be a matter of minutes before I can listen to this thing.

Ask BYacey about this preamp. He has been using his DIY Trans Amp to do all his tracking for a number of years.

Oh, just got a load of Burr Brown opamps from Sounds of Thunder to try out in this bad boy.

Thanks Charlie!!!



cj

 

[SonsOfThunder]

Glad to hear your progress is coming along on this... It might be a great ckt to fit into "another" product (8x mic pre)

Glad you got tha package CJ. The DIP14s (quads) are discontinued for some time now. The AD parts are essentially OP275 that can be used with a single-supply.

Peace!
Charlie

 

[CJ]

OK, I hope Rev 3 will get me somewhere:

I reversed one row of of array transistors and changed the current source, added a spot for a buffer, added some pwr lines and increased the pad size as well as the trace size. What you have on your artwork will shrink a little around the edges, so make your traces and pads a little bigger than scale..
Going to stuff this guy tonite.

cj

 

[PRR]

> Heating up the transistors in the array does not cause as big a shift in current.

Shouldn't cause any real shift in current. That's set by the current sources.

You thermal-couple the array for current balance. With the opamps servoing the input array back to balance, it won't be obvious, but a big mismatch between sides will increase 2nd harmonic distortion.

 

[CJ]

Went thru 50 transistors for the trans-amp. I measure Veb, Vec, and Ic at two different levels: 1ma amd 5 ma, which is hopefully around the current levels in which the transistors will be used.
I found that despite people telling me to match up Veb, matching Ic and Vec was the way to go, because there was very little difference among the transistors as far as Veb. Only a few millivolts. But there was a lot of difference in Ic, which meant I could have a meaningful way to sort the transistors. I recorded the data on paper first:

http://vacuumbrain.com/The_Lab/TA/Valley_People/vp_match2.jpg

Then I transfered them into an Excel spreadsheet. Although this took about an hour, it probably save me 5 hours trying to match them up. I had 6 columns of data. With the spreadsheet, I could add up all the values and divide by 50 to get an avarage for each of the six columns. Then, I could sort the columns in ascending order and screen out the varoius paramters that were furthest from the average, one at a time. Here is some of the data in unsorted form:

Here are the final 8 for the array.
The avg values are shown on the bottom line:

I used double sided tape to hold the transistors:

http://vacuumbrain.com/The_Lab/TA/Valley_People/vp_match1.jpg

I used matched transistors for the current source. There are three of them, so I shaved two of them flat to get better thermal coupling. They are tie wrapped together with some transistor thermal mounting pad material, which does not dry out like white heatsink compound:

I did the same thing for the array, only I did not have to shave them as I designed the board so they could be placed back to back. I paired up the transistors with the closet parameters, even though this might not make much difference, since four are in parallel on each side of the diff amp:

The engineer at work took home his plus/minus supply, right when I was ready to test this thing! So I had to build a new supply which I will use tonite to tweak this thing.

cj

 

[tk@halmi]

Neat. I use the same board idea for sorting and measuring, but I use velcro instead of double sided tape.

 

[bcarso]

Under what conditions are you matching Vce and Ic?

 

[CJ]

Supply= 10vdc, current at 1 ma and at 5 ma, simple bias resistor from base to minus.

 

[SonsOfThunder]

Cj,

Any new gnus? :green:

 

[CJ]

Finally!
Passing some 1 kcps sine wave. 70 db max.

Turns out this diff amp with the +/gnd/- pwr supply does not like signal generators with grounds hooked up to it. doh!

So how did I finally get this transformerless circuit working?
I used a transformer!

That Triad 600:600 on the input to isolate the sig gen. After all, a microphone is floating, and a generator, no always.

Had some motorboating going on. Along with some other weird stuff. Hey, I'm new to this transistor stuff. Tube circuits always have x-formers on the fron end, so I never had to worry about gronding issues with the generator, scope, etc.

Testing with a Linear Devices LT1078. Getting roll off at 12 kcps, so I will plug in one of these sexy B/B 2134's thank you very much and you know who you are.

cj

 

[SonsOfThunder]

Cool Beans CJ!

a microphone is floating, and a generator, no always.

Not that it is good safety practice, but the easy way to fix this is a 2-prong AC adapter with the tab snipped off and stuck on the AC cord of the sig gen.

Testing with a Linear Devices LT1078. Getting roll off at 12 kcps

My first thought was, "oh the OA won't make any diff" until I saw that the slew rate on that OA is 0.07 typical... Just about ANYthing will give you better high freq response. But! that one would keep you out of high freq oscillation on the opamps!! :green: :green: :green:

Keep us informed CJ! Best wishes.

 

[CJ]

OK, here is the "state of the schematic" post.
Right now, this is where I am at.
Sorry if you have to squint. Maybe a printout:

Here's a clear pdf, but it's 2.6 MB for reasons unkown.
http://vacuumbrain.com/The_Lab/TA/Valley_People/vp_cj.pdf


Anyway, the transistors looked like they are matched alright. I had to play with some resistors to get the voltages as close as I could to the schematic on the app notes.

The collector resistors below Q1 and Q2 needed to be tweaked. They affect the collector voltage. Lowering the resistance will bring the collectors down towards the minus rail. This does not seem to affect the emitter/base voltages. Those voltages are set up with the100K bias resistors.

R6, the current source resistor, can be tweaked to help move voltages around. I think I am running around 8 ma with the 8 transistor array. Since I dropped the collector resistors to about 2.2k, the ac gain of the opamp will go up, so I may need to lower the 3.3M feedback resistors.

My question is, now what? What do I do with the output? two caps into a buffer? Then a cap coming off pin 6 of a buffer as my plus out in a singel mode?

I guess I could go into a transformer, but that would defeat the purpose of a transformerless preamp. I can not connect the plus and minus outputs directly to a buffer without coupling caps because of the offset voltage on the outputs of the instrument amps, right?

How about just two caps on the plus and minus outputs to the XLR out? Can I get away without a buffer? I guess it depends on which dual opamp I use. I am going to go with a Burr Brown 2134.

Thanks for any help!

cj

 

[SonsOfThunder]

[quote author="CJ"]My question is, now what? What do I do with the output? I can not connect the plus and minus outputs directly to a buffer without coupling caps because of the offset voltage on the outputs of the instrument amps, right? How about just two caps on the plus and minus outputs to the XLR out?[/quote]
Shove that 2134 on the o/p and measure each out W.R.T ground. C4 and C5 are going to decouple the inputs of the OA. So I think there is every possibility that it only depends on the offset voltage of the OAs themselves.

Can I get away without a buffer?

Yes, I think that this is the point!!

HTH!