I recently bought an Audio Aero DAC, and I would like to change the OPA 2134 I/V stages, the DAC chip is an AD 1853 (balanced outputs).
For more information on this converter:
http://www.audioaero.com/ Products then Prima D/A Converter 24/192 MKII
http://www.anagramtech.com/
http://www.6moons.com/audioreviews/audioaero/prima.html
The sound is very musical, near vinyl, but it lacks a bit of transparency, 4 op-amps and a tube in the signal path!
I wonder if this I/V stage adapted from an old input circuit of a power amplifier design that I tested a few months ago (the project is in standby waiting for a really good source, CD or turn-table, - now I have both
), could be a good challenger…
I/V Schematic
It’s a first simulation design, probably I will go for a folded cascode, for the moment I just focus on the I/V conversion.
Help and comments are welcome.
For more information on this converter:
http://www.audioaero.com/ Products then Prima D/A Converter 24/192 MKII
http://www.anagramtech.com/
http://www.6moons.com/audioreviews/audioaero/prima.html
The sound is very musical, near vinyl, but it lacks a bit of transparency, 4 op-amps and a tube in the signal path!
I wonder if this I/V stage adapted from an old input circuit of a power amplifier design that I tested a few months ago (the project is in standby waiting for a really good source, CD or turn-table, - now I have both
), could be a good challenger…I/V Schematic
It’s a first simulation design, probably I will go for a folded cascode, for the moment I just focus on the I/V conversion.
Help and comments are welcome.
Please use a variable resistor 20 turns for the load resistor. Methinks that you'll need 1.500 ohms for a good output.
At least in my design, with a 3,5 mA current source to the bottom rail ans TDA1541 this was the optimum value to get 0 dB and an undistorted 0dB 1 kHz sine output.
I used a 2SK1078 miniMOSFET and no source resistance. The resistance will give quite some AC effect on the DAC output and the ideal is 0 mV AC IMHO.
At least in my design, with a 3,5 mA current source to the bottom rail ans TDA1541 this was the optimum value to get 0 dB and an undistorted 0dB 1 kHz sine output.
I used a 2SK1078 miniMOSFET and no source resistance. The resistance will give quite some AC effect on the DAC output and the ideal is 0 mV AC IMHO.
Thanks triode_al,
What do you mean by "the load resistor"? The 2 * 22R resistors between the
fets, for dac load ?
I have in mind to use a trim-pot there to adjust the input for the dac at
0V, and try to lower the resistors values to 10R, but as the fets present
a "load" around 30R, I think that it will not change a lot, in my case the
reference is 2.7V for the AD1853, but for simulation it’s easier to
reference the circuit to ground.
Have you a schematic of your design? As well as the SK1078 datasheet, it
seems not to be an easy part to get, does the complementary part exist ?
What do you mean by "the load resistor"? The 2 * 22R resistors between the
fets, for dac load ?
I have in mind to use a trim-pot there to adjust the input for the dac at
0V, and try to lower the resistors values to 10R, but as the fets present
a "load" around 30R, I think that it will not change a lot, in my case the
reference is 2.7V for the AD1853, but for simulation it’s easier to
reference the circuit to ground.
Have you a schematic of your design? As well as the SK1078 datasheet, it
seems not to be an easy part to get, does the complementary part exist ?
This is the folded cascode design :
I/V_Conv_Folded Cascode.
Probably better with cascoding Q7/Q8.
DAC output is 3mA AC peak to peak, for those who are skill in the art of I/V conversion, which current values do you recommend, and what would be their transistors choice?
I didn’t built it yet, I will do the first tests with MPSA06/56 for Q5/Q6-Q7/Q8 biased at around 15 mA, maybe it’s too much and 6/8mA with SC2240/SA970 will be better?
I/V_Conv_Folded Cascode.
Probably better with cascoding Q7/Q8.
DAC output is 3mA AC peak to peak, for those who are skill in the art of I/V conversion, which current values do you recommend, and what would be their transistors choice?
I didn’t built it yet, I will do the first tests with MPSA06/56 for Q5/Q6-Q7/Q8 biased at around 15 mA, maybe it’s too much and 6/8mA with SC2240/SA970 will be better?
Not yet build it, probably this w.e. here just a thd simulation:
I/V thd.
At full scale 6V pk to pk output thd is mostly 2nd at around –80 dB.
At 2V pk to pk 2nd thd is at – 90 dB.
And at 1V pk to pk, -100 dB.
Probably real results will be a little worse.
Input impedance is 11R (2*22R in //) Is it lower enough for the AD 1853 ? I think I could go for half the value with this fets 2*10R and increase their bias current to 4 mA.
If someone is experienced with the AD 1853 advises are welcome.
I/V thd.
At full scale 6V pk to pk output thd is mostly 2nd at around –80 dB.
At 2V pk to pk 2nd thd is at – 90 dB.
And at 1V pk to pk, -100 dB.
Probably real results will be a little worse.
Input impedance is 11R (2*22R in //) Is it lower enough for the AD 1853 ? I think I could go for half the value with this fets 2*10R and increase their bias current to 4 mA.
If someone is experienced with the AD 1853 advises are welcome.
The folding cascode changes things of course.
I have no experience with AD 1853. Only philips 154x. And they have a current source. So the input impedance is crucial, and will make a different current in both halves. I recollect Rbroer has designed I?V cascodes of various types.
BTW have you ever seen the input stage of Le Monstre? There the FETS are in the signal path (not just current sources) while the nice harmonic degradation comes from a real cascode. This seems to me a good desgin direction. I add the file of the original article. The little monster from Jean Hiraga is reknowned.
2SK1078 is a MOSFET in SMD format, very low drain resistance and not loading the DAC-output resistance at all. I loaded the bootom with a current source (2SK30) and this wirks fine. But of course a different design direction! Literally also: it is single ended.
I have no experience with AD 1853. Only philips 154x. And they have a current source. So the input impedance is crucial, and will make a different current in both halves. I recollect Rbroer has designed I?V cascodes of various types.
BTW have you ever seen the input stage of Le Monstre? There the FETS are in the signal path (not just current sources) while the nice harmonic degradation comes from a real cascode. This seems to me a good desgin direction. I add the file of the original article. The little monster from Jean Hiraga is reknowned.
2SK1078 is a MOSFET in SMD format, very low drain resistance and not loading the DAC-output resistance at all. I loaded the bootom with a current source (2SK30) and this wirks fine. But of course a different design direction! Literally also: it is single ended.
Hi, triode_al,
Yes I know the “Le Monstre” I build it a few years ago, it’s a very very good sounding amplifier.
Actually after reading the rbroer threads, I give up (for the moment the CFP, as it seems to be less musical than the simple common base stage).
I am currently working on his design called “Less simple I/V stage”, measures are quite good, I am tuning it, current sources values, probably replacing the current source that feeds the common base and the folded cascode, by a simple resistor.
I will listen it, in a few days after some more tests and tunings, and post here or on his thread the measuring results, then listening feelings.
Could you post a schematic of your design?
Yes I know the “Le Monstre” I build it a few years ago, it’s a very very good sounding amplifier.
Actually after reading the rbroer threads, I give up (for the moment the CFP, as it seems to be less musical than the simple common base stage).
I am currently working on his design called “Less simple I/V stage”, measures are quite good, I am tuning it, current sources values, probably replacing the current source that feeds the common base and the folded cascode, by a simple resistor.
I will listen it, in a few days after some more tests and tunings, and post here or on his thread the measuring results, then listening feelings.
Could you post a schematic of your design?
The input Z will be higher than 22R/2, as you will not have zero ohms looking into the JFET/BJT compound. That figure usually goes down as you up the current. Porbably will have second order dominant, and it will be much more than -80 dB. Could be some cancellation to bring the number down.
Jocko
Jocko
Hi Jocko,
I give up for the moment the CFP complementary design, and start to study your “simple” design, well more precisely the folded cascode design, as rbroer did on his “less simple I/V for TDA1541 thread, as described in my last post.
I report here, the circuit:
Replacing T11 the upper current source by a resistor, give the same measuring results and I think that it will improve sound quality, I will check this in a few days.
Cascoding T5 by fet T9 don’t improve measures, what is your opinion on it?
All bjt’s are 2SA970/2SC2240 biased at 6mA is it ok?
Measuring results (on real circuit) are nice, around -80 db thd 2, -90 db thd 3, at 0 db output, for 3mA pk to pk input, noise is a little less than -130 dB.
I will post my actual schematic tomorrow.
Did you carry some test and/or listenings on “hybrid” CFP fet/bjt, it seems that classic bjt CFP don’t sound “good” or musical, even if measures are much better.
Have you experiment/design with the AD1853 DAC?
Thanks to help us and share your knowledge for us, to learn a little more about musical circuits design.
I give up for the moment the CFP complementary design, and start to study your “simple” design, well more precisely the folded cascode design, as rbroer did on his “less simple I/V for TDA1541 thread, as described in my last post.
I report here, the circuit:
An externally hosted image should be here but it was not working when we last tested it.
Replacing T11 the upper current source by a resistor, give the same measuring results and I think that it will improve sound quality, I will check this in a few days.
Cascoding T5 by fet T9 don’t improve measures, what is your opinion on it?
All bjt’s are 2SA970/2SC2240 biased at 6mA is it ok?
Measuring results (on real circuit) are nice, around -80 db thd 2, -90 db thd 3, at 0 db output, for 3mA pk to pk input, noise is a little less than -130 dB.
I will post my actual schematic tomorrow.
Did you carry some test and/or listenings on “hybrid” CFP fet/bjt, it seems that classic bjt CFP don’t sound “good” or musical, even if measures are much better.
Have you experiment/design with the AD1853 DAC?
Thanks to help us and share your knowledge for us, to learn a little more about musical circuits design.
Here is my actually learning/measuring circuit:
In this shematic DAC is referenced to ground (0V), for easyer simulations.
I will cascode Q1 and post some FFT measures this afternoon.
The DAC outputs must be referenced to 2.7V, is it best to use the AD1853 reference, or never mind, a TL431 or even some leds will be ok?
An externally hosted image should be here but it was not working when we last tested it.
In this shematic DAC is referenced to ground (0V), for easyer simulations.
I will cascode Q1 and post some FFT measures this afternoon.
The DAC outputs must be referenced to 2.7V, is it best to use the AD1853 reference, or never mind, a TL431 or even some leds will be ok?
An externally hosted image should be here but it was not working when we last tested it.
Replacing T11 with a resistor will hurt PSRR. Just ask Charlie Hansen. Cascoding the current source probably will not help measured results.
The numbers sound about right for the version that I came up with.
Never tried your JEFT/BJT hybrid. The CFP can be oscillation prone. It needs a zobel.
The only AD DAC that I used was the AD1862. A very nice part, very ESD sensitive, and pricey. I prefer working with it than any of the BB parts.
Jocko
The numbers sound about right for the version that I came up with.
Never tried your JEFT/BJT hybrid. The CFP can be oscillation prone. It needs a zobel.
The only AD DAC that I used was the AD1862. A very nice part, very ESD sensitive, and pricey. I prefer working with it than any of the BB parts.
Jocko
Hi Dad (?)
Attached find the scheme I am using.
The sound is “liquid” in NOS.
You state: “classic bjt CFP don’t sound “good” or musical, even if measures are much better”, this might have to do with the drain resistance.
Any case, the drain resistance of this 2SK1078 is rather low and that might be a reason why it is OK here.
It can be possible to replace the tube with a transistor like the 2SC1775 (very low Rb) and bias that on 5 V.
2SK1078 Power MOS FET Enhancement Type, SMD ----- Specs:
ON Resistance : RDS (ON) = 0.58 Ω (Typ.) at VGS = 10 V, ID = 0.4 A and > 0,8 Ω around 5 mA.
Vthreshold = typically between 1,4 and 1,7 volt (ID = 2 mA, VDS = 10 V)
Maximums :
Drain−Source Voltage 60 V
Drain−Gate Voltage (RGS = 20 kΩ) 60 V
Gate−Source Voltage 20 V
Drain Current pulse > 1 A
Power Dissipation > 400 mW
I have sampled to within 2 mV Tthreshold as I have some 100 pieces.
(Information from the TD62M4503AFN Toshiba chip with combines 4 of these devices).
I use the same principle of a cascade of FET and tube on my RIAA MC input (2SK240 + E88CC) though there with classic gate input. It is simply marvelous.
To me T9 in the RBROER scheme is to hide T5. This means that there will be no emitter temperature changes at T5 with voltage swing. The Audiophile ‘gang’ – Héphaistos specifically - recommend such measures. Actually I’d use something like that additionally also on T6.
Attached find the scheme I am using.
The sound is “liquid” in NOS.
You state: “classic bjt CFP don’t sound “good” or musical, even if measures are much better”, this might have to do with the drain resistance.
Any case, the drain resistance of this 2SK1078 is rather low and that might be a reason why it is OK here.
It can be possible to replace the tube with a transistor like the 2SC1775 (very low Rb) and bias that on 5 V.
2SK1078 Power MOS FET Enhancement Type, SMD ----- Specs:
ON Resistance : RDS (ON) = 0.58 Ω (Typ.) at VGS = 10 V, ID = 0.4 A and > 0,8 Ω around 5 mA.
Vthreshold = typically between 1,4 and 1,7 volt (ID = 2 mA, VDS = 10 V)
Maximums :
Drain−Source Voltage 60 V
Drain−Gate Voltage (RGS = 20 kΩ) 60 V
Gate−Source Voltage 20 V
Drain Current pulse > 1 A
Power Dissipation > 400 mW
I have sampled to within 2 mV Tthreshold as I have some 100 pieces.
(Information from the TD62M4503AFN Toshiba chip with combines 4 of these devices).
I use the same principle of a cascade of FET and tube on my RIAA MC input (2SK240 + E88CC) though there with classic gate input. It is simply marvelous.
To me T9 in the RBROER scheme is to hide T5. This means that there will be no emitter temperature changes at T5 with voltage swing. The Audiophile ‘gang’ – Héphaistos specifically - recommend such measures. Actually I’d use something like that additionally also on T6.
Attachments
First, some precisions about the circuit:
X1 reference is 6.9V.
X2/X3 references are 2.5V.
Q1/Q2 bias current is 5.5 mA.
Q5 sinks 2.7 mA.
I through R9 is 2.7 mA.
Input current for 0 dB (2V pk to pk) output, is 3mA pk to pk.
Here some measures on the real circuit:
0 dB full scale.
0 dB “thd zoom”.
-10 dB full scale.
-10 dB “thd zoom”.
I tested cascoding Q1 and Q5 (in the last schematic, real thd result (not simulations) are identical within or without cascoding.
I will carry on more tests, with a square wave signal, and imd measures, once again cascoding and not cascoding...
Hi Jocko,
“Replacing T11 with a resistor will hurt PSRR. Just ask Charlie Hansen. Cascoding the current source probably will not help measured results.”
Agree of course about PSRR, but in the case of good power supply or even with a “perfect” power supply, maybe on the musicality/reality of the perceived sound a resistor could be better no?
You are right cascoding help nothing on measures, but what about sounding results ?, I feel that for sound, cascoding both (Q1/Q5) will be nice, do you know Jocko? Or may I ask to Charles Hansen ?
All these questions are because I will draw a pcb with various possibilities, cascoding or not current sources or resistors, I can’t envisage all the possibilities in one board, too complex for well tuning the circuit for all the cases. And then trying/listening these various possibilities.
What would you choose for references, led’s or chips like TL431 if it matter (except reference stability of course) I mean in a “sounding” way?
What about the reference for the DAC best to use the DAC one or doesn’t matter.
I want to try/listen and learn by myself, but some advises are welcome J
Hi triode_al
I have just see your post while I was writing this one, I will read it more in depht.
“Justcallmedad” is just an allusion from a movie called Devil’s Advocate…
The icon is from another movie, Looking for Richard..., so Justcallme Richard.
X1 reference is 6.9V.
X2/X3 references are 2.5V.
Q1/Q2 bias current is 5.5 mA.
Q5 sinks 2.7 mA.
I through R9 is 2.7 mA.
Input current for 0 dB (2V pk to pk) output, is 3mA pk to pk.
Here some measures on the real circuit:
0 dB full scale.
0 dB “thd zoom”.
-10 dB full scale.
-10 dB “thd zoom”.
I tested cascoding Q1 and Q5 (in the last schematic, real thd result (not simulations) are identical within or without cascoding.
I will carry on more tests, with a square wave signal, and imd measures, once again cascoding and not cascoding...
Hi Jocko,
“Replacing T11 with a resistor will hurt PSRR. Just ask Charlie Hansen. Cascoding the current source probably will not help measured results.”
Agree of course about PSRR, but in the case of good power supply or even with a “perfect” power supply, maybe on the musicality/reality of the perceived sound a resistor could be better no?
You are right cascoding help nothing on measures, but what about sounding results ?, I feel that for sound, cascoding both (Q1/Q5) will be nice, do you know Jocko? Or may I ask to Charles Hansen ?
All these questions are because I will draw a pcb with various possibilities, cascoding or not current sources or resistors, I can’t envisage all the possibilities in one board, too complex for well tuning the circuit for all the cases. And then trying/listening these various possibilities.
What would you choose for references, led’s or chips like TL431 if it matter (except reference stability of course) I mean in a “sounding” way?
What about the reference for the DAC best to use the DAC one or doesn’t matter.
I want to try/listen and learn by myself, but some advises are welcome J
Hi triode_al
I have just see your post while I was writing this one, I will read it more in depht.
“Justcallmedad” is just an allusion from a movie called Devil’s Advocate…
The icon is from another movie, Looking for Richard..., so Justcallme Richard.
I can not disclose the details that I know of Charlie's commercial ventures, but I do know that he abandoned the folded cascode w/resistor (instead of CCS) for something that had lots of current mirrors.
One day, don't ask when, I will make a version of mine that has current mirrors everywhere. Just like Charlie's, only different. [Inside joke]
Remember..........most of the time you are hearing the effects of the power supply. As for cascodes in general:
This mainly applies to power amps, but...................
The way that you bypass the voltage divider on the cascode element has a dramatic effect on the sound. Look closely and you will see minor peaking and the like in the response. This is one place that a MOSFET might be a winner, as there is more isolation between the gate and the channel. Provided of course that you employ a gate damping resistor.
Jocko
One day, don't ask when, I will make a version of mine that has current mirrors everywhere. Just like Charlie's, only different. [Inside joke]
Remember..........most of the time you are hearing the effects of the power supply. As for cascodes in general:
This mainly applies to power amps, but...................
The way that you bypass the voltage divider on the cascode element has a dramatic effect on the sound. Look closely and you will see minor peaking and the like in the response. This is one place that a MOSFET might be a winner, as there is more isolation between the gate and the channel. Provided of course that you employ a gate damping resistor.
Jocko
Justcallmedad said:
JCMD,
As your measurements show, no need to go to the CFP
circuit, results are very good. WRT these measurements what is
DAC OP Z used and what is OP voltage.
Bear in mind that the AA module uses AD1853 which only
has from memory +- 0.75mA OP. So load R will have to be
quite a bit higher than 1k to get decent OP swing.
AFAIK AD1853 has open drain type OP and so will be
very high impedance. This will help with linearity of the gnd base
IP BJT.
Good work.
Cheers,
Terry
Hi Jocko,
“I can not disclose the details that I know of Charlie's commercial ventures, but I do know that he abandoned the folded cascode w/resistor (instead of CCS) for something that had lots of current mirrors.
I suppose that the use of current mirrors is also because the design is something like balanced or differential ?
“This is one place that a MOSFET might be a winner, as there is more isolation between the gate and the channel. Provided of course that you employ a gate damping resistor.”
It’s also the case for jfet’s I suppose ? For small currents (2-10 mA) I would use jfet's (2-10 mA), still best isolation between gate and channel no ?
What about using fet’s as current sources/sinks in lieu of bjt’s + led’s, or Zeners or V ref ?
Simplest design and best PSRR ?
Hi Terry,
I use 1k resistor in series with the sine generator to simulate the DAC impedance, it seems to be a current value in DAC’s simulations… apparently it could be much higher than that in the AD1853 (open drain).
Output is 2V pk to pk or 0 db, for 3mA pk to peak input.
Yes +- 0.75 mA is not too much, but for the moment I focus on the I/V conversion and the global design, when I will be satisfied by the design, the measures, and the sound of course, I will redesign the circuit for balanced operation and then set the gain to a correct value.
---------------------------------
This is the new fet cascoded schematic I didn’t change the R1 resistor by a current source, I will try both in the future listening test pcb.
With some new values for the current sources:
Q1/Q2 bias current is 7 mA.
Q1/Q5 bias current is 4.5 mA.
Is the voltage divider for the cascode fet J1 correct Jocko ?
New FFT graphs with the actual values of current sources.
With higher current values for Q1/Q5 we could get substantially best distortion figures, but as I haven’t fet’s that can handle more than 4 mA for a Vgs around 1V (Vce for Q5), I just speak about fet J2 I keep it like this for the moment..
With 8 mA for Q1/Q5 thd figures are much better for high levels.
0dB_Output.
-10dB_Output.
“I can not disclose the details that I know of Charlie's commercial ventures, but I do know that he abandoned the folded cascode w/resistor (instead of CCS) for something that had lots of current mirrors.
I suppose that the use of current mirrors is also because the design is something like balanced or differential ?
“This is one place that a MOSFET might be a winner, as there is more isolation between the gate and the channel. Provided of course that you employ a gate damping resistor.”
It’s also the case for jfet’s I suppose ? For small currents (2-10 mA) I would use jfet's (2-10 mA), still best isolation between gate and channel no ?
What about using fet’s as current sources/sinks in lieu of bjt’s + led’s, or Zeners or V ref ?
Simplest design and best PSRR ?
Hi Terry,
I use 1k resistor in series with the sine generator to simulate the DAC impedance, it seems to be a current value in DAC’s simulations… apparently it could be much higher than that in the AD1853 (open drain).
Output is 2V pk to pk or 0 db, for 3mA pk to peak input.
Yes +- 0.75 mA is not too much, but for the moment I focus on the I/V conversion and the global design, when I will be satisfied by the design, the measures, and the sound of course, I will redesign the circuit for balanced operation and then set the gain to a correct value.
---------------------------------
This is the new fet cascoded schematic I didn’t change the R1 resistor by a current source, I will try both in the future listening test pcb.
With some new values for the current sources:
Q1/Q2 bias current is 7 mA.
Q1/Q5 bias current is 4.5 mA.
An externally hosted image should be here but it was not working when we last tested it.
Is the voltage divider for the cascode fet J1 correct Jocko ?
New FFT graphs with the actual values of current sources.
With higher current values for Q1/Q5 we could get substantially best distortion figures, but as I haven’t fet’s that can handle more than 4 mA for a Vgs around 1V (Vce for Q5), I just speak about fet J2 I keep it like this for the moment..
With 8 mA for Q1/Q5 thd figures are much better for high levels.
0dB_Output.
-10dB_Output.
Is the voltage divider correct???
I dunno........I'm not paying that close attention.
When I was referring to the voltage divider, I was not implying adding an additional cascode stage. Perhaps I should have been more specific. Any voltage divider for a cascode can do strange things. The only real way is to build one, and measure it.
But basically..............if you hang a bypass cap on the divider in a BJT circuit, it can mess up the response. So, you need a really clean supply in order to leave it off. But.......if you do not have one.......like in a power amp for example........you make not like the sound of a folded cascode if you are using a resistor for the current source. Ergo, Charlie comes up with something else. But you seem to have an ok supply. You could try using just a resistive divider for Q1, and leave off the cap. Then for grins, try bypassing it. Or bypass it, and shove a 1K resistor between the base and the divider. Of course, this assumes that you actually have a working unit, and not a simulation. Try messing with stuff like that. You may find it more enlightening than experimenting with different types of CCS.
Jocko
I dunno........I'm not paying that close attention.
When I was referring to the voltage divider, I was not implying adding an additional cascode stage. Perhaps I should have been more specific. Any voltage divider for a cascode can do strange things. The only real way is to build one, and measure it.
But basically..............if you hang a bypass cap on the divider in a BJT circuit, it can mess up the response. So, you need a really clean supply in order to leave it off. But.......if you do not have one.......like in a power amp for example........you make not like the sound of a folded cascode if you are using a resistor for the current source. Ergo, Charlie comes up with something else. But you seem to have an ok supply. You could try using just a resistive divider for Q1, and leave off the cap. Then for grins, try bypassing it. Or bypass it, and shove a 1K resistor between the base and the divider. Of course, this assumes that you actually have a working unit, and not a simulation. Try messing with stuff like that. You may find it more enlightening than experimenting with different types of CCS.
Jocko
A comment on the input configuration:
Q12 and Q2 must have the same current and voltage component, or the Vbe will differ. The assumption is they are equal. So there should be a current source instead of R3 just like that with Q4.
And when you are sourcing anyway, why not start mirroring...
Still tubes are soo simple to use. But I agree, I have a bias on tubes.
You can find a lot on voltage dividers and how to balance noice on www.tubecad.com.
Q12 and Q2 must have the same current and voltage component, or the Vbe will differ. The assumption is they are equal. So there should be a current source instead of R3 just like that with Q4.
And when you are sourcing anyway, why not start mirroring...
Still tubes are soo simple to use. But I agree, I have a bias on tubes.
You can find a lot on voltage dividers and how to balance noice on www.tubecad.com.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.