PNP frequency response issue

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Hello, I am trying to get a pnp transistor to carry an ac signal linearly from a high positive voltage source to a low negative voltage output.
I tried this
pnp-1.png
but this is what I got out of it
pnp version 1.png
The pnp transistor was specifically designed for audio http://www.farnell.com/datasheets/866302.pdf
What could be causing the distortion in the 0-6k frequency range?
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
I'm not trying to amplify, just create a buffer. It's not 150v AC it's 150vdc with an ac signal attached. I need the output to be -24vdc.
 
Last edited:

Brownout

Joined Jan 10, 2012
2,390
I'm not trying to amplify, just create a buffer. It's not 150v AC it's 150vdc with an ac signal attached. I need the output to be -24vdc.
So, you're inputting a 150vDC with an unspecified ac riding on top and you want -24vdc out? What about ac voltage on the output? Isn't your AC source specified as a 150v sine? Is there any reason you need the DC offset in your source? Can you not block the DC and just bias the transistor for the -24vdc out? And are you aware that is your DC output is = to the DC supply, you are in cutoff, and thus you have no amplification?
 

AnalogKid

Joined Aug 1, 2013
10,987
Why not just AC couple the signal? A 400 V capacitor can handle 174 volts of DC shift. Your question is not very clear, but this is what I think is a valid response. If it isn't, please re-phrase your question with more detail.

ak
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
I want the ac voltage at the output to be relatively the same as the input, I don't want any amplification, just transmition. My ac source is specified as 5v sine 150v dc offset.
The DC offset is there because the signal is coming off of the plate of a vacuum tube.
At the end of the day what I am trying to do is replace a capacitor with a pnp transistor and compare the sonic attributes between the two. The -24v is required for the biasing of the next tube in the chain.
 

Brownout

Joined Jan 10, 2012
2,390
If you want -24vDC at the output + 5vAC, then your output will swing from -21.5v and -26.5v. You'll need at least 28VDC supply.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
The AC in the actual circuit will be much more significant, maybe up to 100v. I have a -36v supply from a simple zener shunt and I just noticed that it is dropping to -6v when connected.
If you want -24vDC at the output + 5vAC, then your output will swing from -21.5v and -26.5v. You'll need at least 28VDC supply.
I thought AC and DC were separate, why do I need more DC to create more AC? In the sim and on my oscilloscope the DC swing is a tiny fraction of a volt and the DC on the output mimicks the DC on the base. What am I not understanding?
 

Brownout

Joined Jan 10, 2012
2,390
the DC swing is a tiny fraction of a volt and the DC on the output mimicks the DC on the base. What am I not understanding?
You said the DC requirement at the output is -24v

I need the output to be -24vdc.
Given your own requirements, you need a 5v AC signal impressed on a -24vdc level. If that's not correct, then tell us exactly what you need.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
The input signal from the end-game circuit will be about 160vdc and 100vac. I cannot have 160vdc imposed upon the next stage of the amp so normally I would place a capacitor. Capacitors only degrade the sound as no perfect capacitor exists, the better the capacitor you use, the less it will degrade the signal.
What I want to do is test a mosfet or bjt as a capacitor replacement and test the frequency response in comparison to the capacitor.

This means the input to the transistor will be 160vdc and about 100vac. The output must be -20vish depending on what the servo dictates the tube bias must be to null the DC at the output of the amp, therefore I must buffer the DC voltage and pass the AC voltage.

However I just noticed that my negative dc supply feeding the base of the pnp transistor and therefore the collector as well is becoming more positive as the input on the emitter becomes more positive, I'm using a simple 30v adjustable supply for the input. I haven't gone anywhere near 150v and the input is already poisoning the output, I guess LTspice can't always be trusted. This might put a hamper on the whole plan, maybe there is another configuration I can use, hmm.
 

Brownout

Joined Jan 10, 2012
2,390
So you want to level shift the DC from +160vdc to -20vishDC and pass the 100vac without changing the peak voltage. As far as I know, this is not possible using a transistor alone. You'll need to use capacitors at some point to separate the DC component from the AC component. Since you don't want to use capacitors, I see no path for you to achieve your design goals. Sorry, I can't help you.
 

#12

Joined Nov 30, 2010
18,224
I just designed a couple of circuits to do what you are attempting. (I started in electronics with vacuum tubes and I have designed and built vacuum tube amplifiers.) I don't believe your theory is correct because there are audio amplifiers with like, 0.0001% distortion USING CAPACITORS! I've been all through this with audiophools that can hear to 23 KHz and believe that if a fly lands on an aircraft carrier, it will change the way you hear the engines. Why do you think you can beat that by adding another active stage with zero amplification? The ratio of improvement divided by gain is a negative number.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Say what you want about capacitors, but going from a standard capacitor to a Duelund cast (considered best of the best) capacitor changed the sound of my amp entirely, the improvement was mindblowing, it was not a subtle audiophool difference. I only place value in my own experience rather than believe internet claims so clearly capacitors change the sound. Now I doubt I'm going to create a gimmicky replacement that will be on the level of a Duelund cap but I figured perhaps I can get close.
[QUOTE="#12, post: 953382, member: 105487" I don't believe your theory is correct because there are audio amplifiers with like, 0.0001% distortion USING CAPACITORS[/QUOTE]
I'm pretty sure anything you buy these days is going to have inaudible distortion levels, if THD was the deciding factor of sound I can probably spend $20 at best buy and be set.
The tube amp I have right now after some modding sounds 99% like real life, I can easily mistake myself being at a concert and I often get freaked out because I think someone is in the room when really it is just the music. I had another amp with 0.0016% THD and It did not sound life-like at all and wasn't even all that enjoyable.

After thinking about it more carefully I think Brownout might be right, I didn't consider that the AC amplitude would shrink with the DC. I thought they were kind of magically separate from each other, another day another lesson.
 
Last edited:

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Wait a second, if the AC amplitude shrinks when going from 100vac on 160vdc to -20vdc than why does it not when the servo throws -20vdc on the line after the capacitor? What am I missing?
 

AnalogKid

Joined Aug 1, 2013
10,987
At the end of the day what I am trying to do is replace a capacitor with a pnp transistor and compare the sonic attributes between the two.
Nothing will introduce less distortion than a single high-quality coupling capacitor. And of the various circuits available to level-shift an AC signal on a DC pedestal, a single transistor is the worst.

ak
 

#12

Joined Nov 30, 2010
18,224
If Dueland capacitors are so great, why are you trying to beat them with an amplifier stage that must cause distortion because change in voltage at the base is slightly non-linear compared to current at the emitter? I've done the math. Metal film caps are the best for lowest dielectric absorption. Polypropylene, polystyrene, in the style of Sprague Orange Drops.

I ran your idea in common emitter and common collector configuration and both of them came up with 50% AC signal loss with a negative supply of -174 volts. Less than that voltage, and you lose even more of the AC. More than that and you lose less of your AC, but you always lose some because your negative supply voltage can not be infinity volts.

Three things to consider:
1) The grid resistor for the next stage is part of the load on the AC voltage gain. Maybe that's how you got this idea about a current follower for the plate resistor. How about using a higher current in the first tube so the plate resistor has much less resistance than the next grid resistor? That seems better than using an emitter follower or a cathode follower stage. In order of increasing current: 12AX7, 12AT7, 12AU7.

2) There are ways to DC couple a pair of tubes. I've seen that, but I've never used it.
3) There is also a method of using one tube for a constant current source in the plate circuit so the plate resistance looks like almost infinity. That makes the next grid resistor the only load on the AC signal.

ps, What is this, "servo" of which you speak? Even a partial bit of schematic would help.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
I think this would answer a lot of questions for you
Design3 symboless sym2-1.png
Forgive the layout, I made it a while ago before I understood the standards of readable schematics but you get the general idea, it is a white cathode follower output stage with DC nulled on the output which is done by a DC servo at the grid of the top tube in the WCF.
You can't direct couple the input stage to the output stage because without the servo bias voltage (-20ish volts) there would be DC on the output of the amp.

Now, I've heard from reputable people that transistor followers don't have any sonic impact on the signal so I figured I would try it out. The problem of turning a positive voltage into a negative voltage is annoying though.

As for your comment on AC signal loss, as it stands my amp would probably blow out my headphones at max volume on the low gain setting. I have plenty of gain I can sacrifice if I had to.
 
Last edited:

Thread Starter

coinmaster

Joined Dec 24, 2015
502
To explain it another way, The AC signal comes off of the anode of the input stage. In order for the tube to function, the anode must be high voltage, in this case 160v.

The AC signal is outputted from the anode, into a capacitor, and then into the grid of the next tube.
As per my schematic above, the grid of the next tube must be at -20ish volts so a capacitor is placed between the two tubes to separate the DC.
A DC servo puts -20ish volts onto the grid of the second tube in order to null DC at the output of the amp. So in other words, the AC signal comes from a 160v source into a -20ish volt line into the grid of the next tube without being altered. The output stage is a unity gain so even though the AC goes from 160v to -20v it does not lose amplitude otherwise the gain stage would be pointless, which is why you guys are confusing me with these AC signal loss proclamations.

I've read that using transistors as followers instead of gain stages does not impose any signature on the sound. Hence my attempt to replicate a capacitor using transistors.
 

#12

Joined Nov 30, 2010
18,224
of the various circuits available to level-shift an AC signal on a DC pedestal, a single transistor is the worst.
I've heard from reputable people that transistor followers don't have any sonic impact
These two statements seem to contradict each other.

About this Servo...is that the pots controlling the power supply to the op-amps? What drives the servo? And why would altering the power supply voltage to an op-amp be an improvement over what the op-amp can do when it has free run to operate solely with the inputs and its output controlling the results?
 
Top