Different differential

[PRR]

> I'm trying to adapt this to a +/- 18V power supply

Why?

You already have +48V handy, don't you?

This circuit "stacks" two stages, so it needs lots of supply voltage. As I recall, it was hard enough to get +18dBm with full 48V supply. 36V is 3/4 of 48V. The input stage eats about 15V. 36V-15V= 21V, say 4V losses, 21V-4V= 17V swing, 6V RMS output, +17.8dBm on paper. Might be much less due to ill-defined bias.

If you must run on lower voltage, the next step would be a turn-around, a current-mirror or a folded cascode. More complication. The ONLY good reason to live with +/- supplies is so the output "can" sit at zero volts for direct coupling; in thie case, the output DC level is poorly defined, will generally not be zero, and won't even be a consistent polarity (allowing electrolytic coupling) unless we severely offset it and reduce output swing.

Don't Waltz in the Modern Dance recital.

 

[tk@halmi]

PRR,

I think it is really neat that it runs on +48.
Would it be possible to do a balanced line out with four (or six) transistors as well?
I know I am really greedy, but it would be very interesting.

Thank You,
Tamas

 

[fum]

[quote author="PRR"]> I'm trying to adapt this to a +/- 18V power supply

You already have +48V handy, don't you?


Don't Waltz in the Modern Dance recital.[/quote]

:green: Look, I'm doin the box step.

In this particular instance, I don't have +48 available. But it's certainly easy enough for me to add. Likely easier than modding the circuit. Board/schematic corrections pending......

ju

 

[PRR]

> a balanced line out

Balanced, but single-ended?

Sure. Cut it in the middle of the input stage. Take the top half, invert all polarities, and use it for the bottom half. i.e., from the input stage there is a 1.5K resistor going up to positive and a 1.5K resistor going down to negative. Modify the input bias, add another output pair.

Hmmmm.... no, doesn't that cancel even-order nonlinearity when you get to the next stage's differential input?

At some point you have to get back to basics. Go entirely single-ended, and forego any pretense of "Balanced" operation except via transformers. Or single-ended and all interfaces transformered.

 

[tk@halmi]

Thank You, PRR. That does make sense.

 

[Rossi]

Thanks, PRR! Great circuit, looks very interesting. I love the simplicity and wanna breadboard it as soon as possible, possibly next weekend. I might need a bit more gain than the recommeded 50 dB available. Would this circuit work well with a step up transformer (maybe 1:2 or 1:3) or would it be better to add another gain stage at the output?

 

[tk@halmi]

I think the aim of the circuit was to have a transformerless input that runs on a single +48 volt supply. IMHO a variable gain second stage of your choice would be a great way to relieve the input stage of "gain-strain" and extend the range of gain to anything you need. PRR's circuit can be built for less than $10 including a rotary switch for gain control. And that is good DIY.

 

[PRR]

> work well with a step up transformer (maybe 1:2 or 1:3)

You can mix-n-match any thing you want.

But: if you have a transformer, why do you want a differential input amplifier????

An ideal transformer makes any single-ended (and simpler) amplifier differential-input.

Most real transformers are not wound for differential secondaries: one end of the secondary is "more earthy" than the other for better rejection and more bandwidth.

And transformerless 150Ω design is dominated by noise voltage concern. If you have a step-up transformer, many compromises come out different.

As TK wisely says: this is a toss-off idea. Different, cheap enough to play with, maybe good for something. But to make it "good for everything" is probably going to lead to massive complication and cost, and lose the cheap/cheerful concept.

Build one of the classic Neve designs. You'll get your gain, and "color".

 

[Rossi]

So step up transformer is not really a good idea. You're right, I'm gonna build it as it is and play around with it.

As for genuine vintage color, I just bought another V272 and some V372D, albeit all in need of repair. The V272 I already fixed; I have some experience with those. The V372D is another animal. It looks like it might sound somewhat Neve-ish with all those tantalum caps. But many of them seem to have gone bad, and it's gonna be a real pain in the ass to replace them. The PCB is very crowded; there are two channels including Haufe transformers in one little box.

 

[chrissugar]

PRR,

What do you think about adding an aditional stage, a buffer with let's say 12 dB gain, something like the Jung buffer. This way the mic preamp would work from 20dB up to 62dB without significant distortion and freq response limitations and also it would drive low impedances with larger voltage swing.
Is this a good idea? Advantages, disadvantages?

chrissugar

 

[tk@halmi]

Chris,

You could utilize a Forssell amp you have already built for the second stage. If you have access to a good selection of rotary switches you could use a 2 pole 12 throw switch. The first eight settings using the first pole could set gain on the PRR stage (the second stage remains at 6dB), and the last four settings using the second pole could go from 6dB to 24dB, or something like that. This would give you better headroom compared to a fixed gain second stage. Just an idea.

Cheers,
Tamas

 

[chrissugar]

Yes Tamas,
I know it will work to use the JFET opamp, the only problem is that it can't drive low impedance loads. I would preffer something like the Walt Jung buffer or the Borbely "Super Buffer".
The "Super Buffer" can drive loads of 100ohm without problems and no significant increase of distortion.

chrissugar

 

[tk@halmi]

The first stage is not particularly low distortion to begin with so I would not spend a lot of energy on a super clean and complicated buffer stage. You can use simple or any buffer after the second gain stage as well. How about a GainBloak? It runs on single rail fine and gives both extra gain and low impedance output.

 

[Rossi]

Ah sure, Gainbloak!! That looks like a cool second stage. Same simplicity as PRR's "different differential" - brothers in spirit, as it were.

Can you simply run it on 48V ? What about ground?

 

[edanderson]

tamas-

what about your mojofet version of the forssell opamp, with the zetex mosfet output stage? that can drive low impedance loads, right? ever tried running it on a single rail?

ed

 

[tk@halmi]

Phew, I just flew coast-to-coast and my internal clock is so messed up that I feel like hurling any time now.

I will post a schematic tomorrow of the two stage circuit: PRRs amp in fornt and gainbloak at the end, all running off the 48V rail.

Mojofet? What a great memory you got there Ed! The mojofet never made it beyond the breadboard. The Hitachi output MOSFETs are excellent, but difficult to obtain and expensive too. I am using easy to source ZVN3310/ZVP3310 pairs in a FET opamp (not published) with good results to drive 600 ohm transformers, not the lower impedance ones tho.

 

[PRR]

> The "Super Buffer" can drive loads of 100ohm without problems

My plan, a 1K5 resistor and push-pull emitter followers, has output impedance near 15 ohms, and should drive 100Ω fine.

 

[tk@halmi]

The reason for the extra stage was the desire for more gain and less strain. I am sure the original circuit works for most applications using condenser mics.

Here is the drawing, right out of PSpice. Sorry about the messy part designations. This was a quick jobby and it needs cleaning up. Please let me know if you see glaring problems with it.

I added an emitter follower to PRR's circuit and decoupled the two stages with a capacitor. This allowed me to "center-bias" the GainBloak. I have not built it, but it works according to the simulator. I will have it built by the end of the week.

Thank You,
Tamas

 

[PRR]

> The reason for the extra stage was the desire for more gain and less strain.

But the proposal was the Jung Buffer. This adds no voltage gain. I don't mind if you replace the output (diodes and push-pull BJTs) with the Diamond Buffer, that's "the same but a little better". Voltage gain won't change.

If you go with a GainBloke type thing: you can probably omit my output buffers. The 1K5 impedance can drive about any amplifier input, or a reasonable gain-pot. No reason to have the buffer. If you leave the different-differential stage bias the same, you may be able to DC-couple the next stage.

If you run anything less-clean than a GainBloke, you run the "risk" of cancelling some of the single-ended even-order distortion. In fact the wholistic way to do it is to use a simple PNP common emitter stage (emitter to + supply rail) so its nonlinearity is in-phase with this stage's nonlinearity. That would also let you re-bias the whole input stage for astonishing common-mode voltage.

 

[tk@halmi]

OK, I will get rid of the buffer.

[quote author="PRR"]In fact the wholistic way to do it is to use a simple PNP common emitter stage (emitter to + supply rail) so its nonlinearity is in-phase with this stage's nonlinearity.[/quote]

That sounds simple.

[quote author="PRR"]That would also let you re-bias the whole input stage for astonishing common-mode voltage.[/quote]

Please bear with me as I am getting confused. Which input stage are you referring to?

Thank You,
Tamas