I can't recall the conversion formula between common emitter and common base h-parameters, I have to dig up my textbooks.
BTW do you know the discrete I/V of Pedja Rogic?
http://www.pedjarogic.com/1541a/pdf/discrete_I-V.pdf
I also have a working I/V where the principle is similar to yours, although realized using an input transformer and a tube. The input transformer is transforming down the input impedance of the cathode of the tube, and there is no separate I/V resistor.
BTW do you know the discrete I/V of Pedja Rogic?
http://www.pedjarogic.com/1541a/pdf/discrete_I-V.pdf
I also have a working I/V where the principle is similar to yours, although realized using an input transformer and a tube. The input transformer is transforming down the input impedance of the cathode of the tube, and there is no separate I/V resistor.
I'm sure many of you have read this (now ancient) web page on I/V conversion:
http://members.chello.nl/~m.heijligers/DAChtml/Analogue/IV.html
I always found it interesting that in the end the opinion given was that passive I/V using a simple resistor sounded the best. I would imagine since the discussion in this thread is about an alternate (yet still discrete) means of I/V, there is obviously disagreement. (Yeah, I don't really have a question, just stating something pretty obvious.)
http://members.chello.nl/~m.heijligers/DAChtml/Analogue/IV.html
I always found it interesting that in the end the opinion given was that passive I/V using a simple resistor sounded the best. I would imagine since the discussion in this thread is about an alternate (yet still discrete) means of I/V, there is obviously disagreement. (Yeah, I don't really have a question, just stating something pretty obvious.)
I did see that link a long time ago (I've been skirting around the idea of a discrete (active) IV for months now).
I think the old passive vs active IV argument is probably as old as the NOS vs OS, TDA1541 vs PCM63, etc.
I've no idea which is better personally. I haven't tried passive. Its easy to do so maybe I should and compare the three. Already, I feel that discrete active is better that integrated active.
I've personally stuck with active just so that the DAC sees as low a resistance as possible. Ideally, current sources/sinks like to see a short circuit to ground.
I think the old passive vs active IV argument is probably as old as the NOS vs OS, TDA1541 vs PCM63, etc.
I've no idea which is better personally. I haven't tried passive. Its easy to do so maybe I should and compare the three. Already, I feel that discrete active is better that integrated active.
I've personally stuck with active just so that the DAC sees as low a resistance as possible. Ideally, current sources/sinks like to see a short circuit to ground.
philpoole said:
Well, it would certainly be easier for you to compare with a simple resistor, than for someone (like me) to compare with the (discrete) active alternative. I think that for my ezDAC (which uses passive I/V), someone like Ray (from the Marantz 63/67 thread) may eventually try a discrete output stage - but I'm not sure if he would bypass the onboard resistor I/V.
INCREASE the Current
Increase the current..........
And make it Sziklai or something........
http://www.diyaudio.com/forums/showthread.php?postid=201512#post201512
philpoole said:
Increase the current..........
And make it Sziklai or something........
http://www.diyaudio.com/forums/showthread.php?postid=201512#post201512
philpoole said:
Phil, you can look at my www below, but briefly, it uses PCM1794/8 and a single op amp for differential to single-ended conversion and gain after the I/V resistor. I've actually just ordered a new batch of boards last night which will have silkscreen/soldermask, if you're interested in getting one.
I'm guessing you mean the current into the base? Its crossed my mind.
Using Sziklai would be interesting, but I'm concerned about oscillations - so need to plan that a bit more. I have a lovely discrete headphone amp, using the current sources seen here, and a Sziklai, and at one stage in its development it, lets say it had a wonderful tone. A single, loud tone that is LOL
! It was fine in the end. It just needed some attention in layout and decoupling etc. I guess right now I'm being lazy. However, I'm keen to try it one day.Thanks for the offer, but I'm very happy with my current DAC at the moment. And it looks like you have one of those op amps in the signal path
. Has Ray avoided that with a variant of his CD63 i/v?Thanks again to everyone for their help, advice and ability to suppress tears of laughter due to my incompetence (at least I manage to do it in the name of comedy
). Its much appreciated!Phil
Returning to the input impedance of the common base transistor stage:
The common emitter h-parameters are given by data sheets. h11e is the ratio of base-emitter voltage change for a given base current change. h21e is the collector current change for a given base current change (often called as beta). These parameters are specified in a given operating point (collector-emitter voltage and collector current).
Drawing a diagram of common emitter and common base, the input imedance of the common base transistor is h11b = h11e / h21e.
h11e is tipically 2 to 10 kohm, h21e is 50 to 500. From these values, the input impedance what the DAC sees is around 4 to 200 ohms. If one seeks for low input impedance, one should go for very high beta transistor, or perhaps a darlington. Since h11e is determined by the slope of the base-emitter diode curve, here again low base current Ib is desired, which condition again mandates high beta and/or moderate collector current.
Many handheld DMMs can measure h21, and I would select and match the highest beta transistors from a batch. My personal favourite is the metal-can low-noise BCY59 which is factory selected for ranges of beta and marked with VII, VIII, IX and X accordingly.
The common emitter h-parameters are given by data sheets. h11e is the ratio of base-emitter voltage change for a given base current change. h21e is the collector current change for a given base current change (often called as beta). These parameters are specified in a given operating point (collector-emitter voltage and collector current).
Drawing a diagram of common emitter and common base, the input imedance of the common base transistor is h11b = h11e / h21e.
h11e is tipically 2 to 10 kohm, h21e is 50 to 500. From these values, the input impedance what the DAC sees is around 4 to 200 ohms. If one seeks for low input impedance, one should go for very high beta transistor, or perhaps a darlington. Since h11e is determined by the slope of the base-emitter diode curve, here again low base current Ib is desired, which condition again mandates high beta and/or moderate collector current.
Many handheld DMMs can measure h21, and I would select and match the highest beta transistors from a batch. My personal favourite is the metal-can low-noise BCY59 which is factory selected for ranges of beta and marked with VII, VIII, IX and X accordingly.
Qserra_tico_tico,
I can easily up the current.
Quickly measuring, I can see the voltage at the input - seen by the DAC is 0.29V (not entirely sure how at the moment - because the voltage across the 240ohm resistor is 0.7v - as you'd expect - anyway, that's to be investigated another day). If I recall, the current source is set up to sink just under 3mA. So the input impedance seen that is caused my the current sink is 0.29/0.003 ~= 96 ohms.
Laszlo,
I am assuming h11e is hie and h21e is hfe.
When I built it, I did try and match the two transistors to be used for the comon base stage's hfe. If I recall, they were around 190 - 200. The datasheet has a typical value for hie as 2.2k.
So, the transistor input impedance is 2200/190 = 11 ohms.
Now (if I haven't already in this post ) I'm going to once more demonstrate my ignorance. For DC, isn't the input impedance that offered by the current source (it could have been a humble resistor) in parallel with the transistor input impedance in series with the variable resistance? i.e. 96 || 11?
For AC, the 110uF capacitance knocks the variable resistance out of the equation, which helps.
Have I got the right end of the stick?
Cheers,
Phil
I've gone away and looked at this and understand now.
I can easily up the current.
Quickly measuring, I can see the voltage at the input - seen by the DAC is 0.29V (not entirely sure how at the moment - because the voltage across the 240ohm resistor is 0.7v - as you'd expect - anyway, that's to be investigated another day). If I recall, the current source is set up to sink just under 3mA. So the input impedance seen that is caused my the current sink is 0.29/0.003 ~= 96 ohms.
Laszlo,
I am assuming h11e is hie and h21e is hfe.
When I built it, I did try and match the two transistors to be used for the comon base stage's hfe. If I recall, they were around 190 - 200. The datasheet has a typical value for hie as 2.2k.
So, the transistor input impedance is 2200/190 = 11 ohms.
Now (if I haven't already in this post
) I'm going to once more demonstrate my ignorance. For DC, isn't the input impedance that offered by the current source (it could have been a humble resistor) in parallel with the transistor input impedance in series with the variable resistance? i.e. 96 || 11?For AC, the 110uF capacitance knocks the variable resistance out of the equation, which helps.
Have I got the right end of the stick?
Cheers,
Phil
Phil,
It is not exactly like this. The current source represents infinite impedance per definitum. The collector of Q1 can be at any voltage (well, within a range) and the current still remains constant. The voltage at the emitter of Q3 is defined by its base voltage, adjustable by the variable resistor. The actual resistance of variable resistor does not come in the picture, either.
For finer adjustment I would chose the value of the variable resistor and R6 so that the voltage on C1 should be about 0.8 to 1.0 V, giving a finer adjustment range. Perhaps a multiturn variable resistor would be also a good idea.
It is not exactly like this. The current source represents infinite impedance per definitum. The collector of Q1 can be at any voltage (well, within a range) and the current still remains constant. The voltage at the emitter of Q3 is defined by its base voltage, adjustable by the variable resistor. The actual resistance of variable resistor does not come in the picture, either.
For finer adjustment I would chose the value of the variable resistor and R6 so that the voltage on C1 should be about 0.8 to 1.0 V, giving a finer adjustment range. Perhaps a multiturn variable resistor would be also a good idea.
Well, I've upgraded the common base transistors to BC550C. I had 3 spare, so I found two that both roughly had a h12e of 500.
So that's a marked improvement, but I can't find a figure for h11e for this device, so I hope the input impedance has lowered, but I don't know for sure (yet).
The datasheet does state that, dependent on sample, h12e can be as high as 800. If I buy 20 or so, I might be able to get a pair relatively close with higher beta values.
I also tried increasing the current, but I didn't have any suitable resistances really. I might pursue that further, but at the moment it is sounding superb (it is now also running from its dedicated PSU, LM317, TL431 regulation).
When I get round to it, I'll draw an updated schematic.
Next for me is to think of some output buffering. Probably just an emitter follower, with another current source, maybe a CFP.
So that's a marked improvement, but I can't find a figure for h11e for this device, so I hope the input impedance has lowered, but I don't know for sure (yet).
The datasheet does state that, dependent on sample, h12e can be as high as 800. If I buy 20 or so, I might be able to get a pair relatively close with higher beta values.
I also tried increasing the current, but I didn't have any suitable resistances really. I might pursue that further, but at the moment it is sounding superb (it is now also running from its dedicated PSU, LM317, TL431 regulation).
When I get round to it, I'll draw an updated schematic.
Next for me is to think of some output buffering. Probably just an emitter follower, with another current source, maybe a CFP.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.