If you don't have any offset adjustment in your Aleph, you can modify it by putting in an appropriate size pot in place of the drain resistor on the first of the front end mosfet/jfet. In many schematics this is R7 and connects to V-. If you look at the schematics for Aleph-J you will see the pot in place of the fixed resistor (often 392R I think).
2Picodumbs posted some very interesting findings on temperature stability/coefficient of zeners and leds. If I understand the Aleph circuit (partially), the zener reference at the top feeding from V+ and into the source pins of the front end (long tailed pair) devices (mosfet/jfet) sets the bias but if this moves around with temperature, it could introduce some offset as well.
This makes the variation suggested by RodeoDave in Post #131 especially interesting for you
led string vref aleph
I don't know if RodeoDave ever built it so maybe try a PM to ask if it worked or try it and let us know!
2Picodumbs posted some very interesting findings on temperature stability/coefficient of zeners and leds. If I understand the Aleph circuit (partially), the zener reference at the top feeding from V+ and into the source pins of the front end (long tailed pair) devices (mosfet/jfet) sets the bias but if this moves around with temperature, it could introduce some offset as well.
This makes the variation suggested by RodeoDave in Post #131 especially interesting for you
led string vref aleph
I don't know if RodeoDave ever built it so maybe try a PM to ask if it worked or try it and let us know!
Thanks for reminding me about the drift compensation thing posted by 2picodumbs, but I can't find the thread...
I am not worried about adjusting the DC, rather than the drift from cold amp to warm. What is the experience regarding DC drift with the headphone Alephs out there? How much does the DC change from cold to warm?
Also, what are your thoughts about using the 2SK216 outputs?
I am not worried about adjusting the DC, rather than the drift from cold amp to warm. What is the experience regarding DC drift with the headphone Alephs out there? How much does the DC change from cold to warm?
Also, what are your thoughts about using the 2SK216 outputs?
I thought it might have gotten relocated to the SS forum but it simply got bumped to page 3. So easy to lose gems around here.
Here you go:
2PicoDumbs is not so dumb at all
Here you go:
2PicoDumbs is not so dumb at all
The plan is to replace the CCS Zener with a LED string fed by a 2SK170...
What is the targeted voltage across the LED string? Or, in other words, what LED type do you have in mind?
boyz
input LTP is doing servo job , regarding output DC offset
LTP CCS - once when CCS is good enough , you'll not hear difference between various types ....... and it would not help with DC offset , if you use temp. compensated one
input LTP is doing servo job , regarding output DC offset
LTP CCS - once when CCS is good enough , you'll not hear difference between various types ....... and it would not help with DC offset , if you use temp. compensated one
Yeah LTP, does a very nice job of controlling dc offset.
No need to play with voltage refetences in this area of the amp.
If you wanted to play with it, there is no harm done though, just maybe not as critical as other areas.
No need to play with voltage refetences in this area of the amp.
If you wanted to play with it, there is no harm done though, just maybe not as critical as other areas.
Last edited:
Just for the purposes of considering what the effect might be.
Using a 9.1V zener here, is likely to result in a positive 50mV delta V across the zener (chassis temperature dependent)
1) Consider how much that will change the CCS current by
2) Then divide this current by 2, since current is shared across jfets in ltp
3) Determine change in voltage across R14
Now consider the fact the outputstage is using a degenerated device of much lower transconductance compared with IRFP240/250/140/150 and most likely biased closer to it's own zero temperature coefficient.
The effect here is likely to be less significant, you could simulate the effect of increasing the zener voltage by 50mV and see what happens to dc offset.
Having said all that and if it makes you sleep better you could combine a 2.7V Zener (or green led) with a 6.2V Zener for a cheap virtually zero temp coefficient reference, obviousy adjusting the ccs resistor where necessary to maintain same current.
Using a 9.1V zener here, is likely to result in a positive 50mV delta V across the zener (chassis temperature dependent)
1) Consider how much that will change the CCS current by
2) Then divide this current by 2, since current is shared across jfets in ltp
3) Determine change in voltage across R14
Now consider the fact the outputstage is using a degenerated device of much lower transconductance compared with IRFP240/250/140/150 and most likely biased closer to it's own zero temperature coefficient.
The effect here is likely to be less significant, you could simulate the effect of increasing the zener voltage by 50mV and see what happens to dc offset.
Having said all that and if it makes you sleep better you could combine a 2.7V Zener (or green led) with a 6.2V Zener for a cheap virtually zero temp coefficient reference, obviousy adjusting the ccs resistor where necessary to maintain same current.
Last edited:
The plan is to replace the CCS Zener with a LED string fed by a 2SK170...
I am still trying to wrap my head around this. What current should be running through the LEDs? Which type of 2SK170 should be used (BL or GR)? Which type of (red) LED would work well (part number/example would be cool)?
LTL4231N will provide that kind of voltage drop, this is green not red though. Not that red is any better than green.
5mA is kind of the sweetspot for the leds I tested, but anything above 3mA would be fine.
5mA is kind of the sweetspot for the leds I tested, but anything above 3mA would be fine.
Last edited:
why wasting JFets there?
ring of two is perfect for task, as complete CCS ; if you insist
on tight reference , use LM385 and that's it
observe TempCo of entire shebang - CCS + LTP ..... and it's still nonsignificant for LTP servo function
edit:
pardon moi , I'll say no more - if personal preferences are in question , who am I to dispute that 🙂
ring of two is perfect for task, as complete CCS ; if you insist
on tight reference , use LM385 and that's it
observe TempCo of entire shebang - CCS + LTP ..... and it's still nonsignificant for LTP servo function
edit:
pardon moi , I'll say no more - if personal preferences are in question , who am I to dispute that 🙂
Last edited:
Like Zenmod says LTP does a great job of controlling dc offset.
Leds are better than zeners with regards to noise, and if a negative temperature coefficient is going to improve the dc offset drift it might be worthwhile, but consider the things I mentioned early.
Most likely a voltage reference at the ccs with a zero tempco is going to be best in this circuit.
Circuits like F4, F6, F5 benefit more from having a negative tempco biasing the output stage.
From a noise perspective the leds will be better though.
LEDs are kind of cool though, you know they are working. Hahahaha
Leds are better than zeners with regards to noise, and if a negative temperature coefficient is going to improve the dc offset drift it might be worthwhile, but consider the things I mentioned early.
Most likely a voltage reference at the ccs with a zero tempco is going to be best in this circuit.
Circuits like F4, F6, F5 benefit more from having a negative tempco biasing the output stage.
From a noise perspective the leds will be better though.
LEDs are kind of cool though, you know they are working. Hahahaha
Last edited:
Picodumbs: So a 2SK170 BL would be preferred over a GR?
Thanks rodeodave!
You want around 5mA give or take a couple of mA is fine.
More current is generally better but we want these things to last forever so I say 5mA is kind of a happy region.
If you get a BL then you would apply some degeneration to drop it down to 5mA. You probably want the degeneration since the jfets have a positive temp coefficient, meaning they will draw more current as they heat up, which will then affect the leds.
So yeah if we’re really considering everything a BL with degeneration is probably going to be better than a gr without degeneration in this specific case. How much better? I couldn’t guess off the top of my head, there are a lot of interacting components here.
I am not debating LED vs Zener vs ring of two. Just trying to understand how things work.
I want as low drift as possible. My cans are 16 Ohm, so even a 0.1 VDC drift is not "nothing".
I want as low drift as possible. My cans are 16 Ohm, so even a 0.1 VDC drift is not "nothing".
- Home
- Amplifiers
- Pass Labs
- Aleph H Schematic - Revealed at Last