Hello all,
I gained enough courage to post my first message onto this forum, excluding the introduction. I am not an engineer, so please bear with me even if I am totally wrong 🙂
I have been reading about differences in the sound of 1969 and 1996 versions of the well-known John-Linsley Hood amplifier (JLH for short). Last week I noticed that the 1969 version did not have a low-pass filter in its input. Indeed, JLH added it to the 1996 version since the bandwidth was 'needlessly wide' in his opinion. I will not argue that.
However, this is where things get interesting. He mentioned he expects a source output impedance of 10 kOhms or less. If we look at the schematic at Geoff's site (thanks Geoff for excellent site, it inspired me to build one and served as a 'reference'), particularly the figure 2 at http://www.tcaas.btinternet.co.uk/jlhupdate.htm, we can see the LP filter components, R3 and C2. Even if we assume the source output impedance is near to zero, this filter causes a frequency dependent phase shift. Given a very high output impedance, say 50 kOhms (don't laugh, my Sony X77ES CD-player specs mention 10 kOhms minimum output load for fixed RCA outputs and 50 kOhms for variable RCA outputs), the phase shift is about 17 degrees at 20 kHz. This might introduce blurriness and inferior definition in the imaging. Am I right?
So, I was thinking of removing R2, R3 and C2 from the circuit. Bandwidth will go up, but I fail to see any reason this thing would become oscillatory or anything. I will try this modification, feeding the thing from a real thing (from a CD player instead of 50 Ohm output function generator) and loaded by a cheap speaker, whilst watching the output at the scope and raising bias slowly from zero. The reason I am asking your opinion is that this modification seems too easy and too beneficial to be clever. I do not understand the mechanisms of transient intermodulation distortion very well. Will they become an issue?
Another good thing would be that I could get rid of two ordinary metalfilm resistors and one bulk ceramic 330 pF capacitor in the signal path. I have read that big ceramics are very nonlinear and of no use in the signal path. Does this apply to ceramics in this lowly capacitance range too?
Oh, just to mention, I cannot grab a soldering iron and go on to work just now, since I first have to fix one of the channels. I did a little modification and now bias seems to suddenly jump from one value to another. DC offset tracks the bias jumps as well. Probably a cold solder joint there. The modification was to replace R8, originally 2k7 metalfoil, with a 1621R Vishay-Dale resistor. I lost some gain (about 6 dB) but given the 2 Vrms CD output that was not an issue with JLH's sensitivity. Increased feedback dropped the output impedance from 0.28 Ohms to 0.17 Ohms. Now the amp controls the drums and guitars a lot better. Cymbals are a lot better defined, but I think the quality of the feedback resistor plays some role here. I am quite puzzled, since some people seem to like small feedback (Rudy van Stratum comes into mind first) but my ears tell otherwise 😕
Some background information may be of interest: the amplifier is the 2003 version (actually it was first 1996 version, just added the updates) using 12.5 V rails and Iq of 3.0 A. The power supply features two capacitance multipliers (one per channel). I am driving lowly Sonab speakers with the amplifier. These speakers have 4 Ohms nominal impedance and their minimum impedance seems to become very low (no actual measurements yet). So the small rail voltage is not a real issue yet with these power levels (about 20 watts in 2 Ohms and 15 watts in 4 Ohms). The signal source I am using is the aforementioned Sony player. I have no preamplifier of any kind so I have to drive the amplifier from the 50 kOhm variable output.
So, I think it is now time to dress a flame suit and wait for your reactions 🙂
I gained enough courage to post my first message onto this forum, excluding the introduction. I am not an engineer, so please bear with me even if I am totally wrong 🙂
I have been reading about differences in the sound of 1969 and 1996 versions of the well-known John-Linsley Hood amplifier (JLH for short). Last week I noticed that the 1969 version did not have a low-pass filter in its input. Indeed, JLH added it to the 1996 version since the bandwidth was 'needlessly wide' in his opinion. I will not argue that.
However, this is where things get interesting. He mentioned he expects a source output impedance of 10 kOhms or less. If we look at the schematic at Geoff's site (thanks Geoff for excellent site, it inspired me to build one and served as a 'reference'), particularly the figure 2 at http://www.tcaas.btinternet.co.uk/jlhupdate.htm, we can see the LP filter components, R3 and C2. Even if we assume the source output impedance is near to zero, this filter causes a frequency dependent phase shift. Given a very high output impedance, say 50 kOhms (don't laugh, my Sony X77ES CD-player specs mention 10 kOhms minimum output load for fixed RCA outputs and 50 kOhms for variable RCA outputs), the phase shift is about 17 degrees at 20 kHz. This might introduce blurriness and inferior definition in the imaging. Am I right?
So, I was thinking of removing R2, R3 and C2 from the circuit. Bandwidth will go up, but I fail to see any reason this thing would become oscillatory or anything. I will try this modification, feeding the thing from a real thing (from a CD player instead of 50 Ohm output function generator) and loaded by a cheap speaker, whilst watching the output at the scope and raising bias slowly from zero. The reason I am asking your opinion is that this modification seems too easy and too beneficial to be clever. I do not understand the mechanisms of transient intermodulation distortion very well. Will they become an issue?
Another good thing would be that I could get rid of two ordinary metalfilm resistors and one bulk ceramic 330 pF capacitor in the signal path. I have read that big ceramics are very nonlinear and of no use in the signal path. Does this apply to ceramics in this lowly capacitance range too?
Oh, just to mention, I cannot grab a soldering iron and go on to work just now, since I first have to fix one of the channels. I did a little modification and now bias seems to suddenly jump from one value to another. DC offset tracks the bias jumps as well. Probably a cold solder joint there. The modification was to replace R8, originally 2k7 metalfoil, with a 1621R Vishay-Dale resistor. I lost some gain (about 6 dB) but given the 2 Vrms CD output that was not an issue with JLH's sensitivity. Increased feedback dropped the output impedance from 0.28 Ohms to 0.17 Ohms. Now the amp controls the drums and guitars a lot better. Cymbals are a lot better defined, but I think the quality of the feedback resistor plays some role here. I am quite puzzled, since some people seem to like small feedback (Rudy van Stratum comes into mind first) but my ears tell otherwise 😕
Some background information may be of interest: the amplifier is the 2003 version (actually it was first 1996 version, just added the updates) using 12.5 V rails and Iq of 3.0 A. The power supply features two capacitance multipliers (one per channel). I am driving lowly Sonab speakers with the amplifier. These speakers have 4 Ohms nominal impedance and their minimum impedance seems to become very low (no actual measurements yet). So the small rail voltage is not a real issue yet with these power levels (about 20 watts in 2 Ohms and 15 watts in 4 Ohms). The signal source I am using is the aforementioned Sony player. I have no preamplifier of any kind so I have to drive the amplifier from the 50 kOhm variable output.
So, I think it is now time to dress a flame suit and wait for your reactions 🙂
Hi hlyytine!!hlyytine said:
Your player have a output impedance of about several hundreds ohms...you are confusing the output impedance of the player with the recomended load impedance...they are not the some thing.
I will bet that your player in the fixed output must hava a output impedance of less than 1kOhm. 🙂
You can take the capacitor out of circuit but you must left R2 and R3 in circuit(in the circuit of Fig.2)
R2 bias the input transistor and if take it out of circuit the amp don't work at all
R3 protect from oscilation
Hello Jorge,
your comments are very encouraging. I think I will measure the output impedance of my CD-player soon. Still, I consider the phase shift annoying even if the source output impedance is near zero Ohms. We are trying to get perfect sound here, aren't we? 🙂
Regarding the oscillations et al, I cannot think of any reason the amplifier would not work (neither did my SPICE simulations). Perhaps you took my words too literally? I meant to connect the input DC blocking capacitor directly to the base of the input transistor. Of course the input capacitor could attain some charge but there's always the base-emitter junction of the input transistor and the load impedance to discharge it. Maybe I misunderstood something? 🙂
your comments are very encouraging. I think I will measure the output impedance of my CD-player soon. Still, I consider the phase shift annoying even if the source output impedance is near zero Ohms. We are trying to get perfect sound here, aren't we? 🙂
Regarding the oscillations et al, I cannot think of any reason the amplifier would not work (neither did my SPICE simulations). Perhaps you took my words too literally? I meant to connect the input DC blocking capacitor directly to the base of the input transistor. Of course the input capacitor could attain some charge but there's always the base-emitter junction of the input transistor and the load impedance to discharge it. Maybe I misunderstood something? 🙂
Hmm thinking more of it...
seems like Jorge is right. There's no path for base-emitter current to flow other than R2 and R3. This seems to be essential of the very working of the amplifier at all (it would hit one rail or another without these). I just registered onto this forum to gain some knowledge but I didn't expect to have it in the next, say, six hours 🙂
My SPICE simulation didn't reveal this artifact because I simulated the circuit only for a few seconds. I noticed a little bit steeper rising of the output voltage during a few hundred of milliseconds but thought it was because of current flowing through the input cap and the load. My bad 🙁
Anyhow, removal of the 330 pF capacitor might do wonders.
seems like Jorge is right. There's no path for base-emitter current to flow other than R2 and R3. This seems to be essential of the very working of the amplifier at all (it would hit one rail or another without these). I just registered onto this forum to gain some knowledge but I didn't expect to have it in the next, say, six hours 🙂
My SPICE simulation didn't reveal this artifact because I simulated the circuit only for a few seconds. I noticed a little bit steeper rising of the output voltage during a few hundred of milliseconds but thought it was because of current flowing through the input cap and the load. My bad 🙁
Anyhow, removal of the 330 pF capacitor might do wonders.
Some more comments,
I don't see how R3 protects anybody from oscillating? Also, I am not using the fixed output of my player (due to lack to any preamp). I guess there's a pot across the fixed output and the variable output is taken from the slide of the pot. Therefore the impedance might be considerably bigger than in the fixed outputs.
Actually, all I wanted to say I once thought why almost everybody makes typos on these forums. Now I know, since I have seemingly mistyped your name. I guess one should not write anything on any computer accessible media when having been at the pub 😉
I don't see how R3 protects anybody from oscillating? Also, I am not using the fixed output of my player (due to lack to any preamp). I guess there's a pot across the fixed output and the variable output is taken from the slide of the pot. Therefore the impedance might be considerably bigger than in the fixed outputs.
Actually, all I wanted to say I once thought why almost everybody makes typos on these forums. Now I know, since I have seemingly mistyped your name. I guess one should not write anything on any computer accessible media when having been at the pub 😉
Haha
I guess that what I said about the pub was true....
Seems like I didn't get Jorge's name wrong, I just recalled I typed it "Jorgen" (more like Scandinavic name, isn't it?). Just forgive me, since I'm under moderation and I don't think I can edit my previously posted messages.
I guess that what I said about the pub was true....
Seems like I didn't get Jorge's name wrong, I just recalled I typed it "Jorgen" (more like Scandinavic name, isn't it?). Just forgive me, since I'm under moderation and I don't think I can edit my previously posted messages.
hlyytine said:
Hi hlyytine!
One thing we have in common...the search for the best sound..😉
So in your case for ultimate perfomance:
-R3 can be 1k...but please not less than that.
-use the fixed output of your Cd and put a 10K log. pot at the input of your amp(and physically near the amp )..the output from the pot have high impedance and can not handle well, the capacity of the interconect cable!
Tube_Dude said:
I suppose so 🙂. I've been thinking of adding a DC servo to the amp, but have not yet figured out an elegant solution. I don't want opamps (I will not argue about their effect on sound quality, I just want to keep the amp all-discrete) and finding an elegant all-discrete solution is not easy.
I think you may have point here. The parasitic capacitances might induce oscillation. I think I'll use those 1621R Vishay-Dale RN60D resistors here as well. Fortunately I have got a scope so I don't have to just blindly swap the parts.
That's a great idea! I have yet to measure the actual output impedance. Unfortunately good quality potentiometers cost at least tens of euros and I am not sure whether you can get a pot that matches the quality of the pot in my CD-player. For that price, that is.
But I must admit I'm still sceptical about the actual output impedance of my CD-player. If it really does have lower output impedance, why they ever would have written the specs as they are in the manual
I'm not certain about the high quality pot in your CD...but a pot in the input of a amplifier allways sound best.
See the manual of one of mine CDs ...this one is a 337ESD...
they tell the ideal load impedance and the output impedance, in this case 200 Ohms 😉
See the link
http://cdp101.hp.infoseek.co.jp/cdp337esd/cdp337esd.html
Tube_Dude said:
Perhaps I exaggerated a little, but at least those pots worth an Euro won't do because their resistance balance between individual channels versus the slide position is generally very poor 🙁. Still, your idea about having the volume control after the interconnects sounds sensible to me.
Ok, now that makes sense to me
. Thanks for clarifying. Now I'm feeling a lot better about this load impedance thing. It seems I have to redo my calculations and re-evaluate whether I should remove the LP capacitor 🙂I tracked the bias problem to be one of those counterfeit MJ15003s, seems like it's almost unpossible to obtain genuine MJ15003s.
Removed C3 and substituted R3 with 1.621K Vishay-Dale, one channel OK, the other oscillates when a low-Z source is connected. Upped R3 to 6.8K metalfilm, this time almost rail-to-rail oscillations at several MHz! Took out R3 entirely, no oscillations at all. Some ringing on the output but nothing serious.
How it sounds? Seems like I cannot sleep this night since I have to relisten every disc one more time but I I am definitely not putting C3 and R3 back 😎
Removed C3 and substituted R3 with 1.621K Vishay-Dale, one channel OK, the other oscillates when a low-Z source is connected. Upped R3 to 6.8K metalfilm, this time almost rail-to-rail oscillations at several MHz! Took out R3 entirely, no oscillations at all. Some ringing on the output but nothing serious.
How it sounds? Seems like I cannot sleep this night since I have to relisten every disc one more time but I I am definitely not putting C3 and R3 back 😎
- Status
- Not open for further replies.