bipolar or fet buffer?

[featherpillow]

You actually built it? Good for you!

Thanks! It sure does feel good to experiment and tweak little circuits like this! And I have to thank you all for your help here...you're helping me to work on the weak areas of my electronics knowledge!

Any optimization/tweaking suggestions in the mean time?

 

[bcarso]

Let's get the new schematic and we will have something to refer to.

 

[Svart]

For some reason I completely missed this thread until now. Good work! I second the wish to see the schematics..

 

[featherpillow]

Here's the schematic. I used a 2N3819 for the N-channel FET, and a 2N3906 for the PNP, which was what I happened to have on-hand:

*edited--schematic moved to a later post

 

[bcarso]

Did you omit a coupling C for the trafo on the schematic (say yes)?

 

[featherpillow]

I didn't use any coupling caps in the prototype circuit--no input. no output.

 

[bcarso]

Well, then you are putting appreciable d.c. through the primary of the transformer, limited by the d.c. resistance and the two 33 ohm R's but in general not a good idea. I would go to a good-sized cap from the amp output to the top of the primary.

Also, I don't know why you are using the 33 ohm R's, or why you are using two of them.

 

[featherpillow]

So I should eliminate the 33 ohm resistors and decouple the transformer, correct?

 

[bcarso]

There may be some single resistor that confers benefits when operating with a specific load, or prevents the buffer from oscillating. But one is enough, whatever its value may be. This is an unbalanced drive on the primary side.

You want to capacitively couple the amp output to the transformer primary, so that you block d.c. and pass signals in the audio range. "Decouple" with a cap refers to local filtering to reduce interstage coupling through (typically) the power supplies.

Also you will probably want some kind of load on the secondary, and possibly a Zobel network (series R-C) for optimal transient response. It depends on the transformer and what it will be seeing in use as a load.

 

[Svart]

damn beat to the punch again.. Bcarso, you must LIVE on the forums.. :green:

:guinness:

Hey Featherpillow, Try a cap in the 100uf-470uf range. Nichicon HE are nice.

Not sure I see a reason to keep the R at all..

Check the Zobel on the Jensen paper for a start..

 

[bcarso]

Svart, I have things I'm putting off and generally too much time on my hands lately. I hope the situation will change soon.

 

[Svart]

Tell me about it. that's usually the case for me but lately I've been pinned down with stuff form the bosses. I have 3 deadlines approaching and for some reason I can't "find" time to finish any of them even though I work on them all the time .... i gotta get more organized.. or less drunk.. :green:

good thing the bosses don't know what they are looking at and can't tell that i've been slacking. ah i feel like a loser..

:?

 

[bcarso]

"good thing the bosses don't know what they are looking at..."

I had that advantage at UCLA for 17 years, and boy did I take it. Fortunately I loved what I was doing so they got more than their money's worth, but it just took forever. I am sure that a number of those professors would find it inconceivable that I have learned to make appropriate engineering compromises and actually finish things on schedule. I've even gotten pretty good at estimating the required time to begin with.

 

[featherpillow]

Hey Featherpillow, Try a cap in the 100uf-470uf range. Nichicon HE are nice.

I've not tried the Nichicon HE series, thanks for the tip. I generally use their UPW series, and like them alot.

My math was suggesting that I try a cap in the region of 220-470uf...
As you can tell, my circuit design skills are very novice, and so I very much appreciate the help!

Check the Zobel on the Jensen paper for a start..

Much thanks again! Here I've been trying to find a simple test jig to use in conjunction with my meter in order to measure the inductance of the 600ohm transformer primary (and in the back of my mind, I'm thinking that there must be a simple turns/inductance/impedance ratio or something, isn't there?). Plus, those Zobel transfer functions are giving me a headache just looking at them!

:guinness: :sam: for both of you.

 

[bcarso]

Get hooked up to the scope and drive the circuit with a square wave generator---put something like your anticipated source impedance in series with the generator as well, if possible. Put something like your anticipated load on the output of the transformer. Observe overshoot/ringing on the output. If necessary, put a series R-C network starting from some suggestion above across the output to common and see how it affects ringing. Make changes to improve.

Lather, rinse, repeat...

 

[PRR]

> The spec sheet says 200 ohms O/P impedance, I can't see how this value is set by R4 and R5

They are saying that only the build-out counts. Which implies that the naked amplifier has negligible (<<100Ω) output impedance.

BTW, the other function of the build-out is so Q2 current does not become infinite when it dumps a fast transient into a long cable.

The actual output impedance is the FET dynamic source impedance (1/Gm, often around 500Ω) divided by the current gain of the BJT. Q2 is probably fairly high-Beta, but R8 saps the current gain. As a rough guess, 500/50= 10 ohm output, which is reasonably negligible compared to 200 ohms build-out. The output impedance does not vary much with Q2 Beta: if Beta is high, Q2 Base taps less current off R8, so the working current gain is pretty stable unless Q2 is very low Beta.

The input impedance could be designed for 10Meg or 100Megs. But in guitar-work, anything over 250K or 500K is high enough. Jensen took ~1Meg just to be sure. Too-high just buzzes worse if the cable comes unplugged.

Fussy guitarists might want to reduce the 1,000pFd cap. That's already high for guitar. A smaller cap may damp the highs less, though it will also pickup radio better. A classic 12AX7 input is more like 300pFd than 1,000pFd.

> this circuit seems to function as a phase splitter as well as a buffer...

No. Look again. It is a fake.

 

[featherpillow]

Okay, I'm currently working with this buffer circuit pushing into the transformer. I eliminated the 33 ohm I had on the ground, and added a 220uf cap after the 33 ohm on the output.

I've checked out the Jensen site to see what Zobel network they use on their 600/600's, but they don't seem to be using one in any of the output application schematics. The 150:600 shows an RC of 4.6K and 620pf on the 600 ohm side, so I thought I might use that as a starting point if I see any ringing. I figured I'd sweep a square wave through the audio band and look for any suspicious activity. I should have some results later this evening hopefully.

 

[bcarso]

BTW you may find that your d.c. output is way high, if the pinchoff voltage of the 3819 is large. This will require making the gate bias divider voltage much closer to ground.

 

[featherpillow]

Yes--DC on the output is a little over 5vdc.

 

[bcarso]

Oh--actually that's low then---if you are running this from a single +15V supply. Ideally you may want to be a little above half of the supply.

Reflecting later, if the d.c. on the output is 5V then there is something wrong. The FET should always have a zero-to-somewhat-negative gate-source voltage. With a +15V supply your voltage divider as described sets the FET gate at 7.1V. The output voltage therefore should be more than this.

I assume you measured this with the input open and not tied to some generator referenced to ground, right? You need a coupling cap at the input too to block d.c. of the source. If you were grounding the input then I would believe the 5V d.c. out.