jfets
I have been following this thread with high interest. I believe you ordered the 2sk170's from Mouser. So did I.I received notification from Mouser that their shipment had been delayed. If you email your address I would be happy to send you some of these devices so you can get on with the project.
I have been following this thread with high interest. I believe you ordered the 2sk170's from Mouser. So did I.I received notification from Mouser that their shipment had been delayed. If you email your address I would be happy to send you some of these devices so you can get on with the project.
That's very kind. I have PM'd you.
Yes me too!
Here you go danspy. I recommend using green LEDs as you should be able to get enough range without needing any more in series. I had to use 3 red LEDs in series for my mosfet version, but if I had used green LEDs two would have been enough. Your circuit is already very simple though. I hope this lives up to your expectations
I will be very interested to hear how you think it sounds. This version is much warmer and less crisp than my previous version. I don't know if it is because of the new input or the different transistors and LED based CCS in the VAS.
Your amplifier is probably already very warm sounding with the bootstrap BJT VAS. Hopefully this won't push it over the edge for you - I know you were missing the H3 sound a bit and this one is very H2'ey sounding. If so, maybe a FET and CCS in the VAS will restore the sound to something you like. Anyway, I proably shouldn't put ideas in your head as now you'll hear everything I said and nothing else. Disregard everything I said and enjoy it
You're going to like those acoustic recordings though
Attachments
Last edited:
Wow, a couple of hours of runtime makes difference. It all just seems more settled now.
Just listened to Norah Jones - the fall, all the way through. It was hands down the most natural and unobtrusive reproduction of this recording I have heard on any push pull amplifier.
If I ever build an amplifier with a LTP input again or more than a handful of active devices I'll be very surprised.
Why you would want to use more complex circuits and more active devices to get better distortion figures than this, except as a purely academic exercise, is beyond me. Sound wise it is everything I have been looking for.
Goosebumps aplenty.
Just listened to Norah Jones - the fall, all the way through. It was hands down the most natural and unobtrusive reproduction of this recording I have heard on any push pull amplifier.
If I ever build an amplifier with a LTP input again or more than a handful of active devices I'll be very surprised.
Why you would want to use more complex circuits and more active devices to get better distortion figures than this, except as a purely academic exercise, is beyond me. Sound wise it is everything I have been looking for.
Goosebumps aplenty.
Last edited:
VAS BJT vs. Fet
Hi mike,
I'll try in a sort of round about way:
they're not really, not at least when it comes to the resolution and harmonic distortion distribution of signal/'things', BUT much easier to stabilize in the HF/VHF region...
Any of the proposed Fet's will be much, much faster than any of the do. BJT's and therefore also more difficult to deal with in this respect...
Think BJT - think low impedance current both in and out circuits...
Think any FET's and U have Voltage control in (with parasitic capacitances), but current out and bassically like with tubes (I believe/'been told') High Impedance circuitry - quite a difference and design changing as well.
All the best and may the good ideas and work/fun, progress and evolve
DocO
Hi mike,
I'll try in a sort of round about way:
they're not really, not at least when it comes to the resolution and harmonic distortion distribution of signal/'things', BUT much easier to stabilize in the HF/VHF region...
Any of the proposed Fet's will be much, much faster than any of the do. BJT's and therefore also more difficult to deal with in this respect...
Think BJT - think low impedance current both in and out circuits...
Think any FET's and U have Voltage control in (with parasitic capacitances), but current out and bassically like with tubes (I believe/'been told') High Impedance circuitry - quite a difference and design changing as well.
All the best and may the good ideas and work/fun, progress and evolve
DocO
Last edited:
Mikelm,
A fet may be better for the VAS, and indeed my circuit with the fet VAS sounded wonderful.
I only changed because I was always a little concerned with the dissipation in the ZVP3310A, though it was probably fine. I have no output protection and do not want any semiconductors failing as a result of overheating.
The only devices I had at hand that I felt could comfortably handle this dissipation was a BD140 transistor or IRFP9610 MOSFET. The modern 9610 has a significant non linearity (according to Nelson Pass), so I chose the transistor. If I had better p channel fet available I would have used it.
As for the VAS ccs - I did some extensive experimenting with CCS's in another project and found that fet based CCS's are prone to instability if not implemented properly. I found BJT CCS's to be more forgiving and prefer them for that reason.
A fet may be better for the VAS, and indeed my circuit with the fet VAS sounded wonderful.
I only changed because I was always a little concerned with the dissipation in the ZVP3310A, though it was probably fine. I have no output protection and do not want any semiconductors failing as a result of overheating.
The only devices I had at hand that I felt could comfortably handle this dissipation was a BD140 transistor or IRFP9610 MOSFET. The modern 9610 has a significant non linearity (according to Nelson Pass), so I chose the transistor. If I had better p channel fet available I would have used it.
As for the VAS ccs - I did some extensive experimenting with CCS's in another project and found that fet based CCS's are prone to instability if not implemented properly. I found BJT CCS's to be more forgiving and prefer them for that reason.
Something else to rouse your interest - 100kHz square wave
The BJT VAS has made everything just a little less twitchy. No ringing at all now on the square wave.
Here is the 100kHz square wave at 20V peak to peak into 7R.
No nasties whatsoever....just a little rounding.
What do you think? Typical single ended VAS character - can supply more current than it can sink...and not helped by the greater capacitance of the p channel output fet. I'm sure I could improve that fall time by upping the VAS current, but what's the point?
Wish I could measure THD too...
The BJT VAS has made everything just a little less twitchy. No ringing at all now on the square wave.
Here is the 100kHz square wave at 20V peak to peak into 7R.
No nasties whatsoever....just a little rounding.
What do you think? Typical single ended VAS character - can supply more current than it can sink...and not helped by the greater capacitance of the p channel output fet. I'm sure I could improve that fall time by upping the VAS current, but what's the point?
Wish I could measure THD too...
Attachments
Last edited:
Yes lineup, that is with the filter attached. I might try it without the filter tomorow just out of interest.
SWF,
In LTSpice there are a variety of pmos, but they don't list their current rating, only their gate charge, Vds and Rdson. For VAS applications, I would think the jfet would be more suitable because its transconductance is lower and this would lead to lower loop gain, perhaps 15-20dB lower than a bipolar in this position. OTOH, you could use a pmos with hefty degeneration. And if you ran input and VAS stages at similar currents, with careful device selection since they operate in antiphase it might be possible to extract a bit of distortion cancellation.....
Hugh
In LTSpice there are a variety of pmos, but they don't list their current rating, only their gate charge, Vds and Rdson. For VAS applications, I would think the jfet would be more suitable because its transconductance is lower and this would lead to lower loop gain, perhaps 15-20dB lower than a bipolar in this position. OTOH, you could use a pmos with hefty degeneration. And if you ran input and VAS stages at similar currents, with careful device selection since they operate in antiphase it might be possible to extract a bit of distortion cancellation.....
Hugh
Thanks Hugh,
My fets are on their way I believe. Going to redo everything over the coming days with both fet and BJT VAS types. Thanks for your thoughts. One thing I am surprised no one has mentioned is the linearity of these devices - I thought BJTs were more linear?
Danspy,
For the bias arrangement I posted earlier, I forgot to mention that the one I posted is to allow the use of mosfets or jfets for the input. If you are using a 2SK170 or similar you will not need a positive bias voltage and should only use the lower half of the circuit (remove D1, C2 and R1).
If you don't read this in time it will work fine as is, but it would be preferable to not have any more devices than you need, as the stability and noise will be a bit better. Sorry I forgot to mention it.
My fets are on their way I believe. Going to redo everything over the coming days with both fet and BJT VAS types. Thanks for your thoughts. One thing I am surprised no one has mentioned is the linearity of these devices - I thought BJTs were more linear?
Danspy,
For the bias arrangement I posted earlier, I forgot to mention that the one I posted is to allow the use of mosfets or jfets for the input. If you are using a 2SK170 or similar you will not need a positive bias voltage and should only use the lower half of the circuit (remove D1, C2 and R1).
If you don't read this in time it will work fine as is, but it would be preferable to not have any more devices than you need, as the stability and noise will be a bit better. Sorry I forgot to mention it.
Last edited:
Dear Swordfish,
YES - BJT are more linear, but their inherent characteristics unfortunately more favours odd harmonic distortion, whereas Fet's does more even order and at higher magnitude since the transconductance is quite an order lower that BJT's...
This was, what I believe You where looking for in the 1st place - right?
That's at least how I have managed to interpret the literature I have been plouwing through, regarding the subject(s).
With Best regards and wishes for Your pending Fet arrivals and subsequent hopefully favourable test's
DocO
Last edited:
One reason to prefer the BJT VAS...the oscillating monster
Hi all,
I just completed new build of the amplifier using a fet in the VAS to further evaluate the use of different devices in the VAS.
I used the exact same circuit I posted in post no 336. All component values were the same except the input drain resistor was increased to 1k (as per my earlier fet version), and the VAS BJT was replaced for a ZVP3310A MOSFET with a 221R gate resistor. The CCS uses the LED reference and a BC550 transistor. Other than that all values we the same and construction methods were the same.
Well, the oscillating monster is back, and worse than ever. It won't even settle down into a 7R resistive load. I haven't had time to try any sort of compensation yet, but I suspect it's going to be difficult to tame.
I think this is because of the fet in the vas, or possibly the increased drain resistor in the input stage. I will try dropping the drain resistance tomorrow and see if the reduced gain from the input stage will help. However, my gut feeling is that it is really the fet in the vas that is causing the issues.
There is little doubt in my mind that the amplifier is far more stable with the BJT vas, to the point of not even needing an output zobel or compensation of any kind.
It will be interesting to see what affect the jfet has when I finally get one to put int the input stage. It is possible the vastly lower transconductance will help - the zvn3310a has a much higher transconductance.
Anyway, just an update to let you know what's happening.
DocO, many thanks for your comments above. Nice advice there. And yes, I do like the idea of the even order distortion from the fet...need to get this thing tamed!
Hi all,
I just completed new build of the amplifier using a fet in the VAS to further evaluate the use of different devices in the VAS.
I used the exact same circuit I posted in post no 336. All component values were the same except the input drain resistor was increased to 1k (as per my earlier fet version), and the VAS BJT was replaced for a ZVP3310A MOSFET with a 221R gate resistor. The CCS uses the LED reference and a BC550 transistor. Other than that all values we the same and construction methods were the same.
Well, the oscillating monster is back, and worse than ever. It won't even settle down into a 7R resistive load. I haven't had time to try any sort of compensation yet, but I suspect it's going to be difficult to tame.
I think this is because of the fet in the vas, or possibly the increased drain resistor in the input stage. I will try dropping the drain resistance tomorrow and see if the reduced gain from the input stage will help. However, my gut feeling is that it is really the fet in the vas that is causing the issues.
There is little doubt in my mind that the amplifier is far more stable with the BJT vas, to the point of not even needing an output zobel or compensation of any kind.
It will be interesting to see what affect the jfet has when I finally get one to put int the input stage. It is possible the vastly lower transconductance will help - the zvn3310a has a much higher transconductance.
Anyway, just an update to let you know what's happening.
DocO, many thanks for your comments above. Nice advice there. And yes, I do like the idea of the even order distortion from the fet...need to get this thing tamed!
Last edited:
No . . . that was the 2nd best method
what is describe was the best I found
JLH used this method on his 80W mosfet amp
Input filter
Are you sure your input filter with C2 from post #336 circuit is really effective....
...maybe if your input V3 has a hi output impedance...
Also, with laterals like 2sk1058/j162 I alwasy put a 33pf cap between the drain and gate of the N mosfet to prevent oscillation...
Keep on the good work, it is quite interesting thread so far.
Are you sure your input filter with C2 from post #336 circuit is really effective....
...maybe if your input V3 has a hi output impedance...Also, with laterals like 2sk1058/j162 I alwasy put a 33pf cap between the drain and gate of the N mosfet to prevent oscillation...
Keep on the good work, it is quite interesting thread so far.
Last edited:
Iv'e been studying the OLG in spice for BD140 & ZTX4424 VAS device and the signatures are hugely different.
the FET has masses more OLG - without VAS degeneration still +50db @ 4megHz !
I needed to add a whopping 470 ohms degeneration to the FET to get similar OLG at HF as the BJT.
So it is now easy for me understand why the two versions sound so different and why the Mosfet needs compensation . . .
I will prepare and post shortly
the FET has masses more OLG - without VAS degeneration still +50db @ 4megHz !
I needed to add a whopping 470 ohms degeneration to the FET to get similar OLG at HF as the BJT.
So it is now easy for me understand why the two versions sound so different and why the Mosfet needs compensation . . .
I will prepare and post shortly
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.