This is getting steamy!
Hi Andrew,
"The action of the bypass cap (which is your queried C4) across the emitter resistor of Q3 is to increase the output impedance and effective gain of this current mirror. The Phase-correcting network, C3 and R3, together with the small emitter resistor bypass capacitor C5, adjust the phase-angle of the feedback in the 1 MHz region".
These are the very words of the man, himself.
I am never surprised to see comments about similarities with Sugden amps, as JLH once told me that he did some design consultancy work for Sugden (amongst very many other UK audio companies, many of whom might surprise you), but I don't know anything about Crimson's circuitry, so I cannot pass any valid comment here.
C13 in this circuit is effectively and simply an 'AC bypass' for the Iq setting arrangement. If you look at the design, the signal divides after exiting from Q10, and the lower half (close to the negative line) connects directly to the gate-stoppers of the lower output mosfets.
However, for the upper part (which ideally should be identical) the signal needs to navigate thro' the more complex impedance presented by Q8, RV4, and R11, before it reaches the gate stoppers of the (positive rail) mosfets, and this would very likely have some small effect on this part of the signal which is not shared by the lower half.
C13 merely provides an 'easier route' for this half of the signal, without having any overall effect on the DC path here, which, of course, controls/sets the Iq for the output devices.
R10 and C8 are there to roll-off the NFB before the phase-shift reaches 180 degrees for overall stability, and after my own quite substantial amendments to the design (some of which are shown above) I ended up using 47k (in place of 470k) for R10, but still using 5pF for C8. This is an area for experimentation if one has a 'scope, and my choice here gives an impeccable square-wave response with my speaker loads.
I have even run this version of the amp without any compensation at all for a short while, which proves it's inherent high stability, but after trying many variations of cap and resistor values, I settled on the above for a very clean and sharp-cornered leading-edge square-wave at 10kHz'ish.
This choice of stabilisation, as opposed to the possibly more common cap between the collector & base of the second stage transistor, unfortunately (generally) gives a marginally higher THD figure at HF. However, the chosen method by JLH allows a wider phase margin of stability, and this is perhaps more important in a DIY design, where 'running close to the limit' and where the constructor might use parts of unknown tolerances, types, and quality etc., might spell disaster. This is especially so if the hapless constructor doesn't even know there is a poblem at HF(at least for a while, until possibly he sees some smoke!) because unfortunately he doesn't have the use of a 'scope.
Also, this chosen stability method reduces TID, improves the amp's 'settling time', is smoother-sounding and is more predictable with 'unusual' LS (or 'special' LS cable!) characteristics.
JLH claims (and he never exaggerated, to my knowledge) THD figs. in the region of 0.002% (-94dB) for his 'standard' design, but I have never measured this for myself, so I don't think that THD is problematic even with this style of compensation.
Regards,
Hi Andrew,
"The action of the bypass cap (which is your queried C4) across the emitter resistor of Q3 is to increase the output impedance and effective gain of this current mirror. The Phase-correcting network, C3 and R3, together with the small emitter resistor bypass capacitor C5, adjust the phase-angle of the feedback in the 1 MHz region".
These are the very words of the man, himself.
I am never surprised to see comments about similarities with Sugden amps, as JLH once told me that he did some design consultancy work for Sugden (amongst very many other UK audio companies, many of whom might surprise you), but I don't know anything about Crimson's circuitry, so I cannot pass any valid comment here.
C13 in this circuit is effectively and simply an 'AC bypass' for the Iq setting arrangement. If you look at the design, the signal divides after exiting from Q10, and the lower half (close to the negative line) connects directly to the gate-stoppers of the lower output mosfets.
However, for the upper part (which ideally should be identical) the signal needs to navigate thro' the more complex impedance presented by Q8, RV4, and R11, before it reaches the gate stoppers of the (positive rail) mosfets, and this would very likely have some small effect on this part of the signal which is not shared by the lower half.
C13 merely provides an 'easier route' for this half of the signal, without having any overall effect on the DC path here, which, of course, controls/sets the Iq for the output devices.
R10 and C8 are there to roll-off the NFB before the phase-shift reaches 180 degrees for overall stability, and after my own quite substantial amendments to the design (some of which are shown above) I ended up using 47k (in place of 470k) for R10, but still using 5pF for C8. This is an area for experimentation if one has a 'scope, and my choice here gives an impeccable square-wave response with my speaker loads.
I have even run this version of the amp without any compensation at all for a short while, which proves it's inherent high stability, but after trying many variations of cap and resistor values, I settled on the above for a very clean and sharp-cornered leading-edge square-wave at 10kHz'ish.
This choice of stabilisation, as opposed to the possibly more common cap between the collector & base of the second stage transistor, unfortunately (generally) gives a marginally higher THD figure at HF. However, the chosen method by JLH allows a wider phase margin of stability, and this is perhaps more important in a DIY design, where 'running close to the limit' and where the constructor might use parts of unknown tolerances, types, and quality etc., might spell disaster. This is especially so if the hapless constructor doesn't even know there is a poblem at HF(at least for a while, until possibly he sees some smoke!) because unfortunately he doesn't have the use of a 'scope.
Also, this chosen stability method reduces TID, improves the amp's 'settling time', is smoother-sounding and is more predictable with 'unusual' LS (or 'special' LS cable!) characteristics.
JLH claims (and he never exaggerated, to my knowledge) THD figs. in the region of 0.002% (-94dB) for his 'standard' design, but I have never measured this for myself, so I don't think that THD is problematic even with this style of compensation.
Regards,
hi bobken,
the information you have given is most informative, and has given me food for thought, i will be at the stage soon where i once again lift the lid on my amp and do a few mods, will most likely try without DC adjust pot along with the bal and vol.
any other thoughts or suggestions you may have or remember will be gratfully recieved, many thanks for taking the trouble with your replies
regards
bob
the information you have given is most informative, and has given me food for thought, i will be at the stage soon where i once again lift the lid on my amp and do a few mods, will most likely try without DC adjust pot along with the bal and vol.
any other thoughts or suggestions you may have or remember will be gratfully recieved, many thanks for taking the trouble with your replies
regards
bob
hi Andrew,
thanks for that, i did have a RC snubber on the fridge decided to replace it anyway low and behold it has sorted out the problem.
replaced with one of those all in one types 100n +100R X rated.
this problem has given me the runaround and caused me much hair pulling and speaking of X rated, that is a good description of my launguage during the hair pulling. oh well at lease the fridge got cleaning out in the process.
when i have recovered from this traumactic experience i will get stuck into the amp
regards
bob
thanks for that, i did have a RC snubber on the fridge decided to replace it anyway low and behold it has sorted out the problem.
replaced with one of those all in one types 100n +100R X rated.
this problem has given me the runaround and caused me much hair pulling and speaking of X rated, that is a good description of my launguage during the hair pulling. oh well at lease the fridge got cleaning out in the process.
when i have recovered from this traumactic experience i will get stuck into the amp
regards
bob
hi fanuk.
i have one of these amps in daily use, although a slightly updated version to the one shown in your first post.
differences are :-Q9 changed to BF870 and zd1, zd2 changed to 12 volt types. lots of other mods to regulated psu.
the amp pcb's that i use are the original Hart electronics types. bottom of this page http://www.williamshart.com/jlhampli.htm
see my post 18
regards
bob
i have one of these amps in daily use, although a slightly updated version to the one shown in your first post.
differences are :-Q9 changed to BF870 and zd1, zd2 changed to 12 volt types. lots of other mods to regulated psu.
the amp pcb's that i use are the original Hart electronics types. bottom of this page http://www.williamshart.com/jlhampli.htm
see my post 18
regards
bob
burbeck said:
OK I am intriqued. Where do we start with modifications?
I too have the williams hart PCB's, two for the power amp and two for the regulators (I bought an additional PSU reg board mistakenly, but turned out to be a good decision as we will see later).
Yeah I have got the VN1210D's working in place of the VN1210M's the original circuit used, and on the same PCB. I now also have ten of the original VN1210M's 1 watters too!.
Supertex's VN1310 is a much better device, better than the IRF510 also, at that transconductance but voltage is limited to 100v max. It was a 80watt design anyway, and has regulator's for the front end, so you can get away with these.
Ciss Crss Coss
VN1310 - 35pf 5pf 15pf
VN1210 - 125pf 20pf 50pf
(alledgely the same for even the TO-220 (D version) compared with the T-O92/T-O39 ; but this seems generalized and hence wrong - the packaging must have some affect surely)
The VN1310 has single nanosecond ton & toff switching speeds. Great Device! IMHO. But now obsolete.
I think prioritising low capacitance is what matters.
Actually, the Mosfet used for the VAS improves the negative rail supply line rejection compared with a BJT, and allows better tweaking of the current mirror preceding it - due to it feeding a higher Zin. I also think the mosfet VAS changes the whole sound of the amplifier.
I suspect much better current mirrors could be used too. With higher Zout - which may further improve negative PSRR. Also a VN1310N3 (in T-O39) package (lower voltage 60v however) or a JFET even, could be cascoded given excellent performance.
To me at least, there seems little point cascoding a jfet or small signal mosfet with a bipolar as the bipolar does much of the voltage gain! A jfet/mosfet or jfet/jfet combo seems more logical! - at least if you want the sound of fets/mosfets.....
There was a good post or "gem" by the user "X-Pro" about a triode type connection. He referenced the book the "Art of linear electronics" about this. And done some investigations himself. By using a certain way of cascoding jfets he could get very low distortion. a Jfet cascoded with a mosfet would be a most interesting VAS to try soundwise. Not something I have seen done often.
Yeah a weakness of the design, is it's low transconductance of the input stage. I plan to increase this to 6mA, use 100r degen resistors. I have heard that some people found the design suscepticle to to the front stage PSU regulators and I think this is why. Also the lag generated over the gate of the VAS connects directly to the negative supply rail. I do not know how this compares with a miller compensation scheme but an ultra low noise negative supply would be preferred here.
The main improvement I have made is using dual lateral's (250w) in T-O3 package - which you can use with the existing PCB & L bracket. The heatsinking has to be improved. By doing this and keeping the rails voltage at 80w @ 8r you make a substantial power increase into low impedance loads. One PSU is required per amp PCB however.
How sensitive is your PSU to tripping ? due to DC offset.
Let us know what mods you have done to your PSU ?
Best Regards
Kevin
hi Kevin,
thanks for that, i will be looking into replacing the VAS MOS FET and increasing the I, to at least 40mA. I wonder if someone has a few of the vn1210d's or the vn1310's left in stock ill do a bit of digging. your cascode idea might also be worth persuing.
is the amp still stable with the changes that you have made? as i have always found it to be absolutly rock solid.
the changes made to the regulators followed the progression of mods and improvements by JLH himself, from the original power darlingtons through to the change to power MOSFET's, then a couple of further changes/mods which consisted of a change of values of the 2 zeners on the gate of series reg MOSFETS from 4v7 to 10v along with their parallel resistors changed from 4k7 to 10k.
with regard to the sensitivity to tripping the psu with offset i have no problems, there was a JLH mod to reduce the sensitivity to tripping with high volumes, cosisting of adding 2 resistors 470k across the 2u2 capacitors in the trip circuit.
if you have the hart pcb's i would say all the above mods were already implimented.
i did have a problem with the psu going into trip at switch on, with all versions of the regulated supply, the mods that i did was to cure this problem. i could let you have details of this when i dig out the paperwork if this is of any interest.
i have built quite a few of these amps and always found them to have a 'nice sound'
regards
bob
thanks for that, i will be looking into replacing the VAS MOS FET and increasing the I, to at least 40mA. I wonder if someone has a few of the vn1210d's or the vn1310's left in stock ill do a bit of digging. your cascode idea might also be worth persuing.
is the amp still stable with the changes that you have made? as i have always found it to be absolutly rock solid.
the changes made to the regulators followed the progression of mods and improvements by JLH himself, from the original power darlingtons through to the change to power MOSFET's, then a couple of further changes/mods which consisted of a change of values of the 2 zeners on the gate of series reg MOSFETS from 4v7 to 10v along with their parallel resistors changed from 4k7 to 10k.
with regard to the sensitivity to tripping the psu with offset i have no problems, there was a JLH mod to reduce the sensitivity to tripping with high volumes, cosisting of adding 2 resistors 470k across the 2u2 capacitors in the trip circuit.
if you have the hart pcb's i would say all the above mods were already implimented.
i did have a problem with the psu going into trip at switch on, with all versions of the regulated supply, the mods that i did was to cure this problem. i could let you have details of this when i dig out the paperwork if this is of any interest.
i have built quite a few of these amps and always found them to have a 'nice sound'
regards
bob
hi Kevin
I have had no luck sourcing the mosfet's vn1210D pity, any one know of a good replacement ? is IRF510 my best option here? the dissapation will be a couple of watts or so with increased I.
I am very interested in the cascode idea jfet/mosfet or mosfet/mosfet i have been simulating circuits of this type with this amp and (according to the sims) the distortion has reduced considerably. i have also been looking at the input stage i think improvments can be made here also.
soon i will will try some of these ideas on the real thing, time being a limiting factor for me.
i do realise that sims do not by any means tell the whole story but can be a usefull tool.
let me know if you have any further ideas on this
regards
bob
I have had no luck sourcing the mosfet's vn1210D pity, any one know of a good replacement ? is IRF510 my best option here? the dissapation will be a couple of watts or so with increased I.
I am very interested in the cascode idea jfet/mosfet or mosfet/mosfet i have been simulating circuits of this type with this amp and (according to the sims) the distortion has reduced considerably. i have also been looking at the input stage i think improvments can be made here also.
soon i will will try some of these ideas on the real thing, time being a limiting factor for me.
i do realise that sims do not by any means tell the whole story but can be a usefull tool.
let me know if you have any further ideas on this
regards
bob
hi,
with regard to specification of the mosfet used as the VAS in this amplifier. the idea is to increase the I through it. increasing the dissipation to 2 watts. which the original cant handle, the logical subtitute is the vn1210D part (unobtainable).
vn1210D, the basic spec / important parameters
Vdds 100v
ID 1.5A
PD 45 W
Ciss 125pf
Coss 50p
Crss 20p
the IRF510 or the IRF610 are about the closest i could find, does anyone have any thoughts on likely sub? bearing in mind it is a VAS transistor.
regards
bob
with regard to specification of the mosfet used as the VAS in this amplifier. the idea is to increase the I through it. increasing the dissipation to 2 watts. which the original cant handle, the logical subtitute is the vn1210D part (unobtainable).
vn1210D, the basic spec / important parameters
Vdds 100v
ID 1.5A
PD 45 W
Ciss 125pf
Coss 50p
Crss 20p
the IRF510 or the IRF610 are about the closest i could find, does anyone have any thoughts on likely sub? bearing in mind it is a VAS transistor.
regards
bob
burbeck said:
IRF510 might be good here, yes.
( I prefer to use IRF510 before IRF610 if I can )
for lower currents it should be slightly better, if data is somewhat as told in Datasheet
Another great upgrade for IRF610 is 2SK216
This is REALLY better than IRF610 and also IRF510!
Search forum for '2SK216' + 'SEWA' + Tyimo
Tyimo has replaced, upgraded his BOZ ( pre + voltage amplifier by Nelson Pass )
with 2SK216.
He is very pleased with results and SOUND.
See my reference Link, at the bottom of this Article
Also 2SK216, not too low price,
if you can find them have considerably better data in some parameters,
than the cheap ordinary HEX IRF610 / IRF510.
2SK216 is a Lateral MOSFET, of another type
and often more suitable for Audio stages, than HEXFET.
For output stages, can be different: Here HEXFETs have some quality
that makes them work very well
( otherwise they would not have been used by Our Grand MOSFET Master Nelson Pass )
---
DiyAudio Reference:
www.passdiy.com - BOZ Single Ended Preamplifier with 2SK216
Regards
lineup
Lineup Audio Labs
cheers lineup, 2sk216 look good. i even have the model in simulator that will keep me quite for awile.
Kevin
did you have any more thoughts about cascode connection for the VAS? i looked at the ref in jlh's "Art of Linear Electronics" this as you say jfet/mosfet or jfet/jfet looks interesting.
off to score a few 2sk216's
regards
bob
Kevin
did you have any more thoughts about cascode connection for the VAS? i looked at the ref in jlh's "Art of Linear Electronics" this as you say jfet/mosfet or jfet/jfet looks interesting.
off to score a few 2sk216's
regards
bob
burbeck said:
Hello Bob,
Long time no hear! Hope you and yours is doing OK!
Just wondering if you got the 2SK216 lateral types and wondered what they sounded like. In fact I think they are preferable to the VN1210D's.
You mention up north the supply tripping etc was the mod you done to pre load the supply with a resistor or add Cdom? Have noticed some printing errors in some of the schematics etc - doesn't help.
Do you know what the pole frequency's are etc approx with the design. This is an area I struggle with. I Know the maths for a Self type design with miller. This design is abit tricker. It's just you mentioned that you have simulated the design. Me and Ltspice don't see eye to eye!
Anyway I plan to get working on mine with various mods in the next week or so and will keep you informed if you want.
Best Regards
Kevin
hi Kevin,
i did manage to source the 2SK216's and a few other bits ready to lift the lid and try out a few mods, however i have not been able to get around to it as yet, due to illlness and work constraints.
i have simed the original circuit and quite a few other ideas, the uprated I in the VAS, also a few cascode VAS variations.
i was side tracked by an idea for a VAS stage by Alex Nikitin (X-Pro) on this site, the thread is called ' diyAudio Forums - VAS with CD-CB stage.
i did a variation (in sims) on the JLH 80Watt using this idea and the results looked very good, 10 fold reduction in distortion.
i realise that sims do not tell the full story, but is an interesting exersize non the less, i would love to get round to trying a few of these ideas asap.
i am very interested in the results that you get with your amp mods and would be very gratful if you would keep me informed of the progress and findings.
the PSU tripping on switch on was cured with a couple of diodes if you want details i will dig out the paper work for you.
i did manage to source the 2SK216's and a few other bits ready to lift the lid and try out a few mods, however i have not been able to get around to it as yet, due to illlness and work constraints.
i have simed the original circuit and quite a few other ideas, the uprated I in the VAS, also a few cascode VAS variations.
i was side tracked by an idea for a VAS stage by Alex Nikitin (X-Pro) on this site, the thread is called ' diyAudio Forums - VAS with CD-CB stage.
i did a variation (in sims) on the JLH 80Watt using this idea and the results looked very good, 10 fold reduction in distortion.
i realise that sims do not tell the full story, but is an interesting exersize non the less, i would love to get round to trying a few of these ideas asap.
i am very interested in the results that you get with your amp mods and would be very gratful if you would keep me informed of the progress and findings.
the PSU tripping on switch on was cured with a couple of diodes if you want details i will dig out the paper work for you.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.