Has anyone seen this front-end before?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
D1 R4 Q18 is a CCS.
D3 Q17 adds a cascode to the CCS.
It is a cascoded CCS.

D3 defines the Vce across Q18.

Using a red LED, or two diodes, increases the Vce across Q18
This may make no difference to the performance of the CCS, or it may improve the performance of the CCS.
I don't know which.

I am suggesting you look into the change and after informing yourself you decide to leave as is, or change it to a red LED.

Why don't you try it yourself and report back the result?
 
I already have it in Roender's power amp.
As he designed it with LEDs as the voltage references, it works very well.

Have you now accepted that it is not a current mirror (CM) and it is actually a constant current source/sink (CCS)?

My last attempt, this is from Sergio Santos, comparison between two CCS type (from the link you did not want to look at), one with two transistors in a feedback configuration and other with red LED transistor and current mirror.
I did not say that what Roender used is not good, just different. Open your eyes.
 

Attachments

  • Sergio's CCS.png
    Sergio's CCS.png
    41.9 KB · Views: 307
Those are not the same as in post1193.

I have opened my eyes and looked at your schematic.
I saw the cascoded CCS and recognised it as almost identical to Roender's, but with a diode for the second stage voltage drop instead of using a second LED.

Oh yes it is the same. When the link start to open, you have to wait a bit and it will show post #55 in that thread, don't start to scroll immediately.
 
I'm sorry, I have trouble understanding how this mechanism would work, could you perhaps illustrate how does 'this nonlinear signal' leaks through the TIS exactly?

Provided the LTP is properly linearized and the TIS isn't loaded with a too low impedance one would expect the TIS to generate the bulk of the distortion.

In a common-emitter TIS, Early effect causes output voltage to affect Vbe and Ib even if Ic doesn't change at all. A BC550C typically has a -60db Vce-to-Vbe coefficient, so for instance a 1V change in Vce will cause Vbe to change 1mV despite Ic being constant. So the collector voltage signal "leaks" into the Vbe, even if the TIS is not under current load (however it is under voltage load).

If the signal at the collector is distorted then, Vbe will also reflect this distorted signal. So for instance if you use a beta-enhanced CE, you will eliminate the distortions "leaked" through Ib, but you will not do anything to fix the distortions leaked through Early effect.

Cascoding the BC550C with another BC550C, the first transistor will have Vce change by only 1mV, so it's Vbe will change by 1uV. This almost always reduces Vbe leakage to far below the rest of the distortion mechanisms in the amplifier.

At this point the main distortion mechanism will probably be OPS input current, which is lowered for instance by buffering the OPS and cascoding the drivers. Eventually your OPS will have low enough input current that the Ib of your TIS cascode transistors will be larger. The distorted voltage at the OPS input will cause similarly distorted current in the Ib of the cascodes and any other capacitances or loads at the TIS, and these load currents will flow through the VAS carrying their distortions into the input stage.

The Baxandall feeds the TIS cascode Ib back into the TIS output node so that it and the distortions in it no longer load the TIS and IPS.

I hope this is clear.
 
Andrew, I don't think Dadod disagrees with you, I think he was referring to a different circuit.

In any case, a diode and transistor (functions like a current mirror or multiplier/divider depending on Vbe/Vf mismatch) can be either a CCS or a VAS/TIS. If it is a signal stage, then it is a TIS. If not, it is a CCS.
 
My last attempt, this is from Sergio Santos, comparison between two CCS type (from the link you did not want to look at), one with two transistors in a feedback configuration and other with red LED transistor and current mirror.
I did not say that what Roender used is not good, just different. Open your eyes.

Hi Damir

In the feedback CCS a diode should be placed on the emitter of Q1, this will negate the temperature affects.
 
That will double the tempco, after all you've added an extra silicon junction.

But on that note, when is it actually desirable in an amp to have a zero-tempco current source? The load of the current source always has it's own tempco after all, so you may actually end up making the overall behavior worse.
 
Hi Andrew
I looked at the schematic and also expected it would benefit from more V across the transistor. But I tried it in simulation and it doesn't seem to.
This is a surprise to me, perhaps you could sim it and we can both try to understand it better? Maybe in a different thread so as not to hijack Edmond's.



Hi Damir
I don't really follow your explanation but the circuit seems to work well anyway;) thanks for the pointer.

Best wishes
David

Hi David,
This was explained and simed by smms73(Sergio) in other thread I pointed to Andrew with the link in the post #1195. I was just a messenger, I did not try to explain how it works.
BR Damir
 
I looked at previous posts. Dadod, a diode and BJT are only a current mirror if the diode is across the B-E of the transistor, and any extra degeneration is equal for both junctions. Otherwise, it does not function as a current mirror. In your CCS the current through the diode does not change the current through the BJT, so it cannot be a current mirror.
 
Last edited:
But on that note, when is it actually desirable in an amp to have a zero-tempco current source? The load of the current source always has it's own tempco after all, so you may actually end up making the overall behavior worse.

Here is an example where a CCS should have a tempco, because the circuit it feeds also has a tempco. I find that this is usually the case for power amplifiers, rather than the CCS tempco causing problems.

http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble.html#post4127099
 
Most of us have an inbuilt reflex to worry about transistors with less than 1Vce. I've found however in my Kmultipliers that for transistors without an obvious quasi-saturation region, they behave well right down to Vcesat. So more than 0.6Vce on a cascoded BC5xx won't do much good.

In a cascoded CCS it is actually the C-B leakage of the cascode transistor that is dominant in the output impedance. So the best way to get higher impedance is to choose a very high-gain cascode transistor (if Ib is low then B-C leakage will also be low) with low Cob.
 
I looked at previous posts. Dadod, a diode and BJT are only a current mirror if the diode is across the B-E of the transistor, and any extra degeneration is equal for both junctions. Otherwise, it does not function as a current mirror. In your CCS the current through the diode does not change the current through the BJT, so it cannot be a current mirror.

If you think that a current mirror should have equal current in both legs than you are right. There current mirrors with unequal current with different degeneration ratio, why not with other transistor and LED in its emitters?
 
This has more to do with semantics than anything. By technicalities the bottom diode and transistor could be considered a current mirror. But I would never use it as a current mirror because it would be a very bad one. I think this is true for most of us, which is why I don't consider it a current mirror.

The top diode and transistor don't function as a current mirror, only as a cascode and bias generator, because the Vf of the diode does not change the Vbe of the transistor and so cannot have any direct current mirror effect. The bottom transistor and diode act as a very nonlinear current mirror with a ratio that rises logarithmically with current. I don't consider it a current mirror because it is not linear and can't really be used as one.
 
This has more to do with semantics than anything. By technicalities the bottom diode and transistor could be considered a current mirror. But I would never use it as a current mirror because it would be a very bad one. I think this is true for most of us, which is why I don't consider it a current mirror.

The top diode and transistor don't function as a current mirror, only as a cascode and bias generator, because the Vf of the diode does not change the Vbe of the transistor and so cannot have any direct current mirror effect. The bottom transistor and diode act as a very nonlinear current mirror with a ratio that rises logarithmically with current. I don't consider it a current mirror because it is not linear and can't really be used as one.

I really don't know why we discus this kind of CCS, which by the way is a good one. I showed an IPS with a "super-VAS" how you called it and hoped some discussions about suitable OPS for it.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.