john_ellis said:
Hello John,
A few questions for you on your JLH design.
Have you tried a CCS load for the phase splitter with 2SC3281 outputs? Have heard this arrangement can work but mostly with a bootstrap load.
Does the 47pF caps you mention over the feedback resistor reduce the OLG bandwidth under 100Khz ? I was also wondering if creating a lag over the base of the phase splitter would get round this. ?
I have heard of one person use even faster Sanken RET transistors with Ft/GBP nearly double the 2SC3281 but large cap 80pF I think.
On a another note. Emitter degeneration. Being a singleton input without any way to cancel the second harmonic like a normal ltp can. One option is to introduce a degen resistor like 100r/200r to maybe cut it by a tenth. A difficulty for me is the current through the feedback resistors etc.
I can double the CCS current to the emitter but the feedback resistor has me stumped. Could be much more than 100uA's. Not sure on that one.
I think i will start off with MJ21194 NPNs with 10r base stoppers and a very tight layout before trying the compensation/faster stuff.
Help appreciatted
Kevin
Hi Kevin
The input stage of JLH uses a low value of emitter resistor already!
I tend to prefer 100 ohm decoupling resistor and 1k feedback instead of 220 ohm and 2.2 k (or was it 2.7k JLH used?)
This resistor limits the OLG and so using 100 ohms instead of 220 gives a useful increase in gain and approx. halves the distortion.
47 pF across 1kohm does not affect performance at 100kHz much (3dB frequency point is 300kHz).
I don't think you would need base resistors with MJ21194's. I think a small feedback capacitor is needed. The original JLH shows a peak in response around a few MHz without any feedback capacitor, and this may make the design sensitive to layout.
cheers
John
The input stage of JLH uses a low value of emitter resistor already!
I tend to prefer 100 ohm decoupling resistor and 1k feedback instead of 220 ohm and 2.2 k (or was it 2.7k JLH used?)
This resistor limits the OLG and so using 100 ohms instead of 220 gives a useful increase in gain and approx. halves the distortion.
47 pF across 1kohm does not affect performance at 100kHz much (3dB frequency point is 300kHz).
I don't think you would need base resistors with MJ21194's. I think a small feedback capacitor is needed. The original JLH shows a peak in response around a few MHz without any feedback capacitor, and this may make the design sensitive to layout.
cheers
John
JPeitzman said:
I'm not surprised your search proved fruitless. Not one of the hundreds of JLH constructors who have contacted me has even mentioned this add-on.
This is perhaps understandable given the comment in the original article, quote:
"If the output impedance of the pre-amplifier is more than a few thousand ohms, the input stage of the amplifier should be modified to include a simple f.e.t. source follower circuit, as shown in Fig. 8. This increases the harmonic distortion to about 0.12%, and is therefore (theoretically) a less attractive solution than a better pre-amplifier."
Now if Tr4 were to be replaced with a JFet it could be a different matter
Geoff
True I did read that too in the original article, though is this the figure with the give FET? I am sure that the increase in distortion would be a different matter with modern JFET's, but I may be mistaken, it wouldn't be the first time .
You did bring up a point that I had considered too, the replacement of tr4 with a JFET, has this been tried. I have seen the PLH, interesting but not quite what I had in mind. I like the idea of a JLH with a high input impedance and some of that JFET character, but still a JLH Class-A through and through.
BTW has anyone tried Zetex transistors in their JLH. I'm doing some rebuilding and modifying and was considering trying these in addition to some Toshiba's and some various others. I am going to be ordering some for some N. Pass projects and thought I might give them a shot in my transistor rolling quest...can you tell I have tubes in my blood
Cheers!
James
. You did bring up a point that I had considered too, the replacement of tr4 with a JFET, has this been tried. I have seen the PLH, interesting but not quite what I had in mind. I like the idea of a JLH with a high input impedance and some of that JFET character, but still a JLH Class-A through and through.
BTW has anyone tried Zetex transistors in their JLH. I'm doing some rebuilding and modifying and was considering trying these in addition to some Toshiba's and some various others. I am going to be ordering some for some N. Pass projects and thought I might give them a shot in my transistor rolling quest...can you tell I have tubes in my blood
Cheers!
James
For your interest, I have built, in addition to the bipolar transistor version of the JLH, a version of a bridged JLH with all FETs and crossed feedback as in Nelson Pass's XA amplifiers :
http://www.diyaudio.com/forums/showthread.php?postid=722428#post722428
The latest version, which is my current reference, has JFET input diff pair (2SJ109BL) with 910R drain resistors. It is easy enough to re-engine this back to the single-ended version of the updated JLH as described in Geoff's excellent website, using JFETs for the first stage and the phase splitter (including the current sources), and MOSFETs for the output stage.
Patrick
http://www.diyaudio.com/forums/showthread.php?postid=722428#post722428
The latest version, which is my current reference, has JFET input diff pair (2SJ109BL) with 910R drain resistors. It is easy enough to re-engine this back to the single-ended version of the updated JLH as described in Geoff's excellent website, using JFETs for the first stage and the phase splitter (including the current sources), and MOSFETs for the output stage.
Patrick
update on the help cry
I think you are right about being near the SOA for the 2N3055. As my cap multiplier on the right hand channel was broken it was running on + and - 26 Volts and even higher on start-up. (the rails sag a little on full load).
I've added a 47 pF across the feedback resistor. This gave very little difference on DC offset, so I think it wasn't oscillating after all.
I repaired my cap. multiplier and I'm on + and - 23.5 Volts rails with a small Ig of 0.75 A. I expect them to sag a little when I increase the Iq.
Now I am doubting how to continue. I happen to have 4 spare 2N3055, so I can double output transistors, increase Iq (rails sag a little more) and hope that I'm safe (else I blow 4 trannies in one go!).
On the other hand I can put in the the 2sc3281 I bought for this purpose a while ago (I didn't dare to risk them before problems were solved.). But then I have to solve the oscillation problem which is a little beyond my knowledge and sets me up with a whole new set of pitfalls.
Advice is welcome!
MArco
I think you are right about being near the SOA for the 2N3055. As my cap multiplier on the right hand channel was broken it was running on + and - 26 Volts and even higher on start-up. (the rails sag a little on full load).
I've added a 47 pF across the feedback resistor. This gave very little difference on DC offset, so I think it wasn't oscillating after all.
I repaired my cap. multiplier and I'm on + and - 23.5 Volts rails with a small Ig of 0.75 A. I expect them to sag a little when I increase the Iq.
Now I am doubting how to continue. I happen to have 4 spare 2N3055, so I can double output transistors, increase Iq (rails sag a little more) and hope that I'm safe (else I blow 4 trannies in one go!).
On the other hand I can put in the the 2sc3281 I bought for this purpose a while ago (I didn't dare to risk them before problems were solved.). But then I have to solve the oscillation problem which is a little beyond my knowledge and sets me up with a whole new set of pitfalls.
Advice is welcome!
MArco
john_ellis said:
Thanks for the reply and informative answer John.
So would you say that the degen resistors of the CE transistor could be biased by a negative potential with a split rail supply? (not used to that being honest unless of diff. polarity!)
I thought at first that you mean't the internal Rb of the transistor in question. As you will know this affects noise.
PS. I am currently just starting my HNC in electronics.
Second Question,
I was thinking of using two different transistors for the OP stage of the JLH. I read briefly what you said on the issues of bootstrapping and the upper transistor speed etc. - positive feedback plus using a diode etc
So I was thinking of using a MJ3281A on top and a MJ21194 on the bottom. Both these are T-O3's. Would be interesting to see what happens with devices with broadly the same gains but different FT's(GBP) are mixed.
The gain linearity of the MJ21194 isn't that bad really.
To be honest I think we could exploit some affects of the output stage!.
I will be building your PLIL design very soon and testing it out on electrostats (seriously low Z at HF) and floorstanders (very low Z at LF) etc and will let you know my findings.
Will use the BC transistors etc.
Best Regards
Kevin
update on heeeeeeeelp
Hello,
I while ago I posted a help cry about the repeated dying of the right channel of my amp.
I've got some really useful replies and I concluded that my right channel output transistors were operating just outside their SOA due to a broken capacitance multiplier supply on the right hand channel leaving me with a higher supply rail voltage.
So what I did:
repaired the capacitance multiplier, and added some extra capacitance on the negative rail of the amplifier board (1220 uF in total after the 1000 uF output cap of the cap multiplier).
Doubled output transistors to 4 per channel, I used 2sc3281 for that, and I've put a 47 pF parallel with the feedback resistor to counteract potential oscillation problems.
I've increased the Iq to 1.5A (on 24V rails) and planning to up it to 2A.
Result: the amps sounds very very nice and it's actually playing. Now I'm praying that it will live for a while. Temp of the heatsinks is about 35 C after a whole day of playing music (15 C above ambient). So that is running nice and cool. The transformers run on almost 50 C so that's not to bad either. The small heat sink for the bridge rectifiers is also on about 50 C.
One thing I'm wondering about: will I actually hear if the amp is oscillating or do I need a scope to find out? Will I have sonic benefits in decreasing capacitance parallel to the feedback resistor?
Thank you all.
Marco
Hello,
I while ago I posted a help cry about the repeated dying of the right channel of my amp.
I've got some really useful replies and I concluded that my right channel output transistors were operating just outside their SOA due to a broken capacitance multiplier supply on the right hand channel leaving me with a higher supply rail voltage.
So what I did:
repaired the capacitance multiplier, and added some extra capacitance on the negative rail of the amplifier board (1220 uF in total after the 1000 uF output cap of the cap multiplier).
Doubled output transistors to 4 per channel, I used 2sc3281 for that, and I've put a 47 pF parallel with the feedback resistor to counteract potential oscillation problems.
I've increased the Iq to 1.5A (on 24V rails) and planning to up it to 2A.
Result: the amps sounds very very nice and it's actually playing. Now I'm praying that it will live for a while. Temp of the heatsinks is about 35 C after a whole day of playing music (15 C above ambient). So that is running nice and cool. The transformers run on almost 50 C so that's not to bad either. The small heat sink for the bridge rectifiers is also on about 50 C.
One thing I'm wondering about: will I actually hear if the amp is oscillating or do I need a scope to find out? Will I have sonic benefits in decreasing capacitance parallel to the feedback resistor?
Thank you all.
Marco
Hi Marco,
You can have some low level VHF oscilation without ever hearing any difference.
for me I think the benefit came for using fast output transistors that required a small compensation cap. I never heard the amp without the cap because it would have been oscillating ( High level ) so it is not possible for me to comment. However, JLH always emphasised that a more stable amp tends sounds smoother than one tending towards instablilty. ( personally I am not sure this applies if the potential oscillation / ringing is as high frequency as it is on a typical simple class A amp )
mike
You can have some low level VHF oscilation without ever hearing any difference.
for me I think the benefit came for using fast output transistors that required a small compensation cap. I never heard the amp without the cap because it would have been oscillating ( High level ) so it is not possible for me to comment. However, JLH always emphasised that a more stable amp tends sounds smoother than one tending towards instablilty. ( personally I am not sure this applies if the potential oscillation / ringing is as high frequency as it is on a typical simple class A amp )
mike
I haven't built this yet, but it simulates very well at only 10 V supply. Look at the 100 kHz square wave and bandwidth below, not slow.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Hello,
Nice circuit I like it.
Yup I have known for a time that adding a DC servo (opamp) for the input current source was the way to go (rather than the CCS arrangement on earlier threads when the NFB cap is taken out). Same as some power amps without global NFB. Having said that it is for headphones, so a high quality film cap on the output shouldn't be THD problem.
Want to try something similiar with a power amp version. PS. There is some tremendous transistors available at low voltage including dual monolithic matched super beta types etc.
Cheers
Kevin
Nice circuit I like it.
Yup I have known for a time that adding a DC servo (opamp) for the input current source was the way to go (rather than the CCS arrangement on earlier threads when the NFB cap is taken out). Same as some power amps without global NFB. Having said that it is for headphones, so a high quality film cap on the output shouldn't be THD problem.
Want to try something similiar with a power amp version. PS. There is some tremendous transistors available at low voltage including dual monolithic matched super beta types etc.
Cheers
Kevin
nelsonvandal said: