I had a multiplier in a valve circuit using bipolar transistors. It showed signs of instability. I replaced it with a FET version that was better. It however needed work to make it reliable. It also gave a little second harmonic distortion. Having said that it made the amplifier viable.
The CRC idea seems likely. It was argued we should use cheap low grade capacitors as first C then high grade oversized to decouple the power amplifier. The wiring the R. That was class ab. We debated it to death. I could see no harm in it and often some good. What it could do is help the multiplier. If so a 0.1R after it? As it's class a no transient power reduction by doing it.
The CRC idea seems likely. It was argued we should use cheap low grade capacitors as first C then high grade oversized to decouple the power amplifier. The wiring the R. That was class ab. We debated it to death. I could see no harm in it and often some good. What it could do is help the multiplier. If so a 0.1R after it? As it's class a no transient power reduction by doing it.
all off my jlh69 amp have an 0.1 to 0.33ohm 10w resistor between first and second stage off psu capacitor .
the first stage have a basic quality capacitors 4.7k uF to 8.8k uF and the second stage have a good quality capacitors never less than 15k uF.
the first stage is decoupled by 1/100 MKP capacitor but not the second .
the first stage have a basic quality capacitors 4.7k uF to 8.8k uF and the second stage have a good quality capacitors never less than 15k uF.
the first stage is decoupled by 1/100 MKP capacitor but not the second .
Hi John
I am sure I replied to this last night, but the post hasn't appeared. Anyway, at 1 watt output, the distortion is about 0.0065 with the test set in circuit and about 0.009 without. The ratio is about the same throughout the power scale.
It is fundamentally not a problem since I am sure I cannot hear any difference, but the change intrigues me.
Regards
Mike
My first post in this massive thread!
I got to page 13 so far, and realized there are over 5000 posts, so I decided to try a shortcut
I have a pair of ebay JLH1969 boards on order from Zerozone V18 PNP Sanken 2SA1216 JLH1969 Class A Power Amp Kit 10W+10W Amplifier Kit | eBay
I have a 'case' with decent heat sinks and two 18V 5A laptop power bricks from a 'interrupted' Aleph Mini project, so this is my starting point. I was adviced to try the JLH, so here I am
Is there somebody with experience from these zerozone kits? Known problems with these? I'm trying to educate myself a bit on these before they arrive.
From reading here so far, I saw a recommendation to convert the 1969-version to DC:
Converting to DC https://www.diyaudio.com/forums/solid-state/3075-jlh-10-watt-class-amplifier-12.html#post27994
There was also some links in the beginning of the thread that no longer works. I have however found this that seems to cover a lot: The Class-A Amplifier Site
I got to page 13 so far, and realized there are over 5000 posts, so I decided to try a shortcut
I have a pair of ebay JLH1969 boards on order from Zerozone V18 PNP Sanken 2SA1216 JLH1969 Class A Power Amp Kit 10W+10W Amplifier Kit | eBay
I have a 'case' with decent heat sinks and two 18V 5A laptop power bricks from a 'interrupted' Aleph Mini project, so this is my starting point. I was adviced to try the JLH, so here I am
Is there somebody with experience from these zerozone kits? Known problems with these? I'm trying to educate myself a bit on these before they arrive.
From reading here so far, I saw a recommendation to convert the 1969-version to DC:
Converting to DC https://www.diyaudio.com/forums/solid-state/3075-jlh-10-watt-class-amplifier-12.html#post27994
There was also some links in the beginning of the thread that no longer works. I have however found this that seems to cover a lot: The Class-A Amplifier Site
Here is a preamp I built for JLH and other uses. It just about works at 5V and is very OK at 9V. As shown 1.3 Vrms in JLH and 1 Vrms into 4K7. I didn't find the original circuit alas. I have changed a few things as it was rather different and I guess for valve devices. The 57 uF is 47 uF + 10 uF. The 10 uF is a high grade. The BC559C was due to not having any BC 327-40. It has almost a valve sound with very low distortion. An easy build on tag strip. Hiss is very low and voice quality notable. Input is 47K. If you have an iphone try feeding via the 57uF. 2K2 is nothing when usual load is 32R. It could for various reasons sound better. Capacitor values are to show depth in the music ( nearer DC than 20 Hz ). That should make comparison with op amps easier. The sound is very soft. However the output is extended above 200 kHz. The softness is niceness only. The circuit mimics a triode valve in some ways.
Seems a bit funny that they say PNP, as does the spec for the sankens, but the JLH 1969 uses NPN.. I guess calling it a JLH would be wrong, or are there variants? I have not even been able to find a schematic for the kit..
The plan is to use it for 4ohm tweeters in an active setup (active filters), so my thinking was that 18V should be ok, just based on the original article https://sound-au.com/tcaas/jlh1969.pdf ? But who knows what this kit is, maybe something completely diffrent than a JLH!
Maybe Iq might need adjustment with the lower voltage, they specify 15-18VAC for the kit. However, not so many watts are needed for my setup, so it could be ok as it is too. Maybe I want to drive 8ohms in the future, then the voltage should probably be higher.
The plan is to use this 6-channel preamp/volume control: Assembeld HIFI 5.1 Remote Volume Controller / 6 Channel Preamp Board | eBay
This should be able to drive the amp, but maybe sound quality is not the best. Not so many affordable options when it comes to 6 channel volume controls..
The plan is to use it for 4ohm tweeters in an active setup (active filters), so my thinking was that 18V should be ok, just based on the original article https://sound-au.com/tcaas/jlh1969.pdf ? But who knows what this kit is, maybe something completely diffrent than a JLH!
Maybe Iq might need adjustment with the lower voltage, they specify 15-18VAC for the kit. However, not so many watts are needed for my setup, so it could be ok as it is too. Maybe I want to drive 8ohms in the future, then the voltage should probably be higher.
The plan is to use this 6-channel preamp/volume control: Assembeld HIFI 5.1 Remote Volume Controller / 6 Channel Preamp Board | eBay
This should be able to drive the amp, but maybe sound quality is not the best. Not so many affordable options when it comes to 6 channel volume controls..
Rallyfinnen is right,the original J. L-H. used NPN 2N3055s. Using PNPs seems a bit odd, especially as PNPs are more expensive than NPNs. I built a J. L-H. using Toshiba 2SC5200s,the amp sounded very "bright",not surprising as the 5200s have a much higher ft.than the 3055s. A word of warning: Toshiba and Sanken audio power transistors attract fakers,as you might expect,their performance is vastly interior to the genuine articles. The Class A Amp. Site is a great source of info.,especially,the collected practical experience of constructors.Finally,if you are using switching psus,make sure that output is TOTALLY isolated from the mains,I have had a few semiconductors destroyed!
Could you elaborate on the isolation to mains in the PSU? I know these PSU's by default have the negative DC connected to PE on the mains side.
What other effects may this have except ground loop hum from the input? (I have modified other amps with hum breaking resistors recommended in some amp book for this reason and other ground loops).
I can't see any reason for concern on the output side going to 'floating' speakers.
On the subject of transistors I know these are probably not the real thing, and I might try replacing them later. I read about people recommending MJ15003, so maybe the complementary 004 PNP might be an alternative in this case. On the other hand, I have tried some most likely copied and inferior sankens in other (AB-)amps, and they did sound nicer than for example KEC transistors.
Since this kit seems to be something different than a standard JLH, I'm not sure how much of the mods and experiences from the original that are actually valid.. I should have done some more homework before I clicked the 'buy' button.. Well, it's not the first time, and probably not the last!
I want to try class A on a budget (hence laptop bricks and cheap kits) just to evaluate if the sound pleases me, or if I can even hear the difference, but maybe I just end up buying many bad kits at a higher total cost, and get a disappointing result in the end.. (the Aleph Mini from the same company was no success either).
What other effects may this have except ground loop hum from the input? (I have modified other amps with hum breaking resistors recommended in some amp book for this reason and other ground loops).
I can't see any reason for concern on the output side going to 'floating' speakers.
On the subject of transistors I know these are probably not the real thing, and I might try replacing them later. I read about people recommending MJ15003, so maybe the complementary 004 PNP might be an alternative in this case. On the other hand, I have tried some most likely copied and inferior sankens in other (AB-)amps, and they did sound nicer than for example KEC transistors.
Since this kit seems to be something different than a standard JLH, I'm not sure how much of the mods and experiences from the original that are actually valid.. I should have done some more homework before I clicked the 'buy' button.. Well, it's not the first time, and probably not the last!
I want to try class A on a budget (hence laptop bricks and cheap kits) just to evaluate if the sound pleases me, or if I can even hear the difference, but maybe I just end up buying many bad kits at a higher total cost, and get a disappointing result in the end.. (the Aleph Mini from the same company was no success either).
Here is a preamp I built for JLH and other uses. It just about works at 5V and is very OK at 9V. As shown 1.3 Vrms in JLH and 1 Vrms into 4K7. I didn't find the original circuit alas. I have changed a few things as it was rather different and I guess for valve devices. The 57 uF is 47 uF + 10 uF. The 10 uF is a high grade. The BC559C was due to not having any BC 327-40. It has almost a valve sound with very low distortion. An easy build on tag strip. Hiss is very low and voice quality notable. Input is 47K. If you have an iphone try feeding via the 57uF. 2K2 is nothing when usual load is 32R. It could for various reasons sound better. Capacitor values are to show depth in the music ( nearer DC than 20 Hz ). That should make comparison with op amps easier. The sound is very soft. However the output is extended above 200 kHz. The softness is niceness only. The circuit mimics a triode valve in some ways.
This is pretty similar, although simplified, to Rod Elliott's P37A.
I'm 90% done with a rev 1 pcb of a similar circuit with a different output offset bias configuration and 2x gain modules sandwiching a pot and with dc servos for all 4 gain modules.
We did debate a PNP version about a year ago. It is neither good nor bad. Goodmans Module 80 was that way. Seems it was a germanium design made with silicon 2955s. I have a hunch for tweeters jlh might be overkill. What ever you do use the output capacitor as the crossover. I cross at 6 kHz that way. Not having an active crossovers is an advantage. You can have active filters in the power amplifier. Kenwood did.
The interesting thing is that circuit has a servo action. It continues to work quite well at 3V.
I am using it with a PAM8406 class D amp that looks remarkably like a jlh distortion wise on paper. As crossover distortion isn't the same when class D as AB it can sound like class A. Also power use makes PSU design easier. Just a bit of fun with USB. Could work for a tweeter. 4 ohms would be interesting. I would use a battery pack.
A single capacitor is not enough for my waveguided tweeters.. The passive crossover for them is actually 1st order (one capacitor) at abt 1,7kHz, but also needs impedance correction or they will have an ugly response hump at resonance. So if I want passive, I can use the passive XO I allready have.
I have done active crossover tuning using dsp in PC (EQ APO) with some more massage of the frequency response, delay between drivers etc, and that did improve sound, so I believe this is the way to go. Phase can be altered too, but I so far I have not been so satisfied with manipulations to phase. I think I need to learn and experiment more with that. Now I only experiment with what amp is best for what driver.
Sorry, I think I'm going way off topic!
When it comes to class D, I tried some cheap 'hyped' boards some time ago, but measurements were really terrible on them, and I was not so impressed by the sound either.
Class D is beyond most people to measure. I used a 12 section 860 kHz.buffered passive filter before I could measure a Hypex design. The analyser before that gave errors. The mighty AP test set has an add on unit. That filter gave about a 16 dB reduction at the switching frequency. It also didn't change bandwidth too much. PSU design is important with class D.
https://www.electroschematics.com/wp-content/uploads/2012/03/lm317-audio-amplifier.gif
This is the very most simple way to do class A. An LM317 is basically a LM741 op amp with booster. A 1.25V bandgap sets a reference voltage. Whilst it's a poor place when trying to be an expert it shows a misunderstanding of class A to say it won't work. As no feedback loop is being used to correct switching no bandwith slewing issues. If we had a gain of 3 we might suspect 300 kHz bandwith at a wild guess. As the 317 mearly uses it's op amp to enhance the reference voltage it can be considered a 3 terminal Gausian device ( see Kirchhoff-Gause, sum of currents entering a node, sum of current exiting a node ). This is the same reasoning I used with my two transistor preamp. The NPN-PNP pair act as one as a Gausian device. Gause said we can consider a town for example to be a node. Thus the seemingly impossible can be solved. What the towns people do is not our concern, it's roads in and out. The LM317 is a town. If you look inside it's a city. I dare say an LM833 and pass transistor would be better. That is if stability issues don't show up. If a PNP pass device I would have doubts.
LD1084 is similar if wanting more current. You really don't. 20% efficient which for a tweeter is fine ( I suspect it's less, I was going by Elektor ).
It is just possible for a tweeter this is an excellent idea. There are better versions from Elektor.
Most tweeters only need a two section crossover to run closer to their resonant point. This might be a 1.5kHz. The Swan isodynamic tweeter has none. However they still recomend two section for avoiding heat build up. If used at 4 kHz they can be single section if it suits.
On my speaker I had to reject fancy tweeters as they just didn't suit. I crossed at 6kHz and had the magic I wanted. The dispersion was correct. The main driver can do 6 kHz and 100 Hz ( 20 Hz - 20 dB ). Phase of the tweeter noticable. Both my lady friend and I can hear this well into our 60's although we struggle to hear spot frequencies above 12 kHz. This means wave shape and spot frequency are not the same by a long way. The ear it seems can still infer leading edge speed. And why shouldn't it? This is where people get mixed up with digital. It can do things even when low bit rate very quickly. Even a home made DAC at 8 bit has nano second speed. 1 bit 256 times over sampling is equal to 16 bit more or less. I would suggest we old folks who love music still infer from memory the true sound given the speed of the leading edge is correct.
The tweeters I use are Swan Hi Vi RT1.3 and Monacore DT74-8 ( DT74-4 better for LM317 I think ). Even though on paper the DT74 is a budget device the reality is different. The Swan needs more work to beat the DT74.
I also have a 3 inch titanium drive unit speaker about the size of a house brick. I usually hate them. I had to add a piezo Motorola clone at 13 kHz first order. By doing that the piezo is transformed as is the titanium driver. The sound is much more like a first class BBC speaker now. They were a gift so I had to put my beliefs to one side and force them to work. Never give up with the cheap piezo horns. They can sound wonderful. If they don't you got it wrong. Measurements back this up. Sometimes reflected off the back wall can work. If you get it wrong they ruin sound ( 80% of the time if being careless ). That was you and not them. Mostly that's phase problems and you need a bit of luck.
This is the very most simple way to do class A. An LM317 is basically a LM741 op amp with booster. A 1.25V bandgap sets a reference voltage. Whilst it's a poor place when trying to be an expert it shows a misunderstanding of class A to say it won't work. As no feedback loop is being used to correct switching no bandwith slewing issues. If we had a gain of 3 we might suspect 300 kHz bandwith at a wild guess. As the 317 mearly uses it's op amp to enhance the reference voltage it can be considered a 3 terminal Gausian device ( see Kirchhoff-Gause, sum of currents entering a node, sum of current exiting a node ). This is the same reasoning I used with my two transistor preamp. The NPN-PNP pair act as one as a Gausian device. Gause said we can consider a town for example to be a node. Thus the seemingly impossible can be solved. What the towns people do is not our concern, it's roads in and out. The LM317 is a town. If you look inside it's a city. I dare say an LM833 and pass transistor would be better. That is if stability issues don't show up. If a PNP pass device I would have doubts.
LD1084 is similar if wanting more current. You really don't. 20% efficient which for a tweeter is fine ( I suspect it's less, I was going by Elektor ).
It is just possible for a tweeter this is an excellent idea. There are better versions from Elektor.
Most tweeters only need a two section crossover to run closer to their resonant point. This might be a 1.5kHz. The Swan isodynamic tweeter has none. However they still recomend two section for avoiding heat build up. If used at 4 kHz they can be single section if it suits.
On my speaker I had to reject fancy tweeters as they just didn't suit. I crossed at 6kHz and had the magic I wanted. The dispersion was correct. The main driver can do 6 kHz and 100 Hz ( 20 Hz - 20 dB ). Phase of the tweeter noticable. Both my lady friend and I can hear this well into our 60's although we struggle to hear spot frequencies above 12 kHz. This means wave shape and spot frequency are not the same by a long way. The ear it seems can still infer leading edge speed. And why shouldn't it? This is where people get mixed up with digital. It can do things even when low bit rate very quickly. Even a home made DAC at 8 bit has nano second speed. 1 bit 256 times over sampling is equal to 16 bit more or less. I would suggest we old folks who love music still infer from memory the true sound given the speed of the leading edge is correct.
The tweeters I use are Swan Hi Vi RT1.3 and Monacore DT74-8 ( DT74-4 better for LM317 I think ). Even though on paper the DT74 is a budget device the reality is different. The Swan needs more work to beat the DT74.
I also have a 3 inch titanium drive unit speaker about the size of a house brick. I usually hate them. I had to add a piezo Motorola clone at 13 kHz first order. By doing that the piezo is transformed as is the titanium driver. The sound is much more like a first class BBC speaker now. They were a gift so I had to put my beliefs to one side and force them to work. Never give up with the cheap piezo horns. They can sound wonderful. If they don't you got it wrong. Measurements back this up. Sometimes reflected off the back wall can work. If you get it wrong they ruin sound ( 80% of the time if being careless ). That was you and not them. Mostly that's phase problems and you need a bit of luck.
I'm not sure I understand everything you're saying Nigel.
I'm in my late 40's myself, and hearing is not perfect. However I only use digital source, and I'm pretty sure I will never convert to vinyl. Just too much trouble in my opinion.
I started with audio when CD was new, and have since moved over to have the digital information stored in the computer instead of optical disks. Processing the data before converting it to analog seems logical to me. People convert the JLH's to get rid of the output cap and seem to like what they hear. Analog filters are a bunch of output caps and coils that can cost a lot of money, and does nothing to improve the sound, so I want to get rid of them all together. I believe caps and coils on the high current side (after amplifier) are a lot worse than in the low current signal path. One can also argue that maybe the distortion harmonics from bass notes will not be heard in the tweeter, since there is a separate amp for the tweeter, and that one is not fed low frequency signals. The bass driver will probably dampen these harmonics pretty well because of it's frequency response', even if they are there.
This is my reasoning for going with DSP filters in the PC and separate amps for each speaker driver.
It also gives the freedom to use class AB (or class D) for low frequency, where more power is needed, and AB distortion is low. However, at higher frequencies and lower output power, class AB is generally not great, and that is where I want to try class A.
I hope Alan will come back and tell us some more about the PSU grounding issues.
I'm in my late 40's myself, and hearing is not perfect. However I only use digital source, and I'm pretty sure I will never convert to vinyl. Just too much trouble in my opinion.
I started with audio when CD was new, and have since moved over to have the digital information stored in the computer instead of optical disks. Processing the data before converting it to analog seems logical to me. People convert the JLH's to get rid of the output cap and seem to like what they hear. Analog filters are a bunch of output caps and coils that can cost a lot of money, and does nothing to improve the sound, so I want to get rid of them all together. I believe caps and coils on the high current side (after amplifier) are a lot worse than in the low current signal path. One can also argue that maybe the distortion harmonics from bass notes will not be heard in the tweeter, since there is a separate amp for the tweeter, and that one is not fed low frequency signals. The bass driver will probably dampen these harmonics pretty well because of it's frequency response', even if they are there.
This is my reasoning for going with DSP filters in the PC and separate amps for each speaker driver.
It also gives the freedom to use class AB (or class D) for low frequency, where more power is needed, and AB distortion is low. However, at higher frequencies and lower output power, class AB is generally not great, and that is where I want to try class A.
I hope Alan will come back and tell us some more about the PSU grounding issues.
Last edited:
That little circuit I gave is supurb with digital. The simplicity works. It did burn in which was the caps no doubt. It sounded nothing special for a day.
Power supplies seem to be like the really good wines against OK wine. Really good not always being expensive. For example Cava is a really good wine and holds up against Champagne or British versions ( supurb ). A battery pack is a bottle of Cava. Sometimes batteries are not better than mains power. However as others have said divorcing the mains is always a good idea. A battery pack turns a dreadful charger into pure music. 5V can just about work if not wanting more than 0.707 Vrms. Digital helps as it has no headroom needs. Mains is a Gausian problem. That is keeping the things you don't want in the town. The roads around can be pure and noise free. Not easy to do.
Rod Elliott. What a Star. He and I seldom agree. I don't mind at all because from his stand point he is right and I see that. That is science made simple. Rod in some eyes is like a bottle of beer. That's only because he isn't out to be esoteric. He knows just as much as those guys.
Rods FET amp class AB is one of his best. I like El Cheapo 1 and 2 very much. An American amp that might have been called Corona was very similar. El Cheapo could be made into class A.
Power supplies seem to be like the really good wines against OK wine. Really good not always being expensive. For example Cava is a really good wine and holds up against Champagne or British versions ( supurb ). A battery pack is a bottle of Cava. Sometimes batteries are not better than mains power. However as others have said divorcing the mains is always a good idea. A battery pack turns a dreadful charger into pure music. 5V can just about work if not wanting more than 0.707 Vrms. Digital helps as it has no headroom needs. Mains is a Gausian problem. That is keeping the things you don't want in the town. The roads around can be pure and noise free. Not easy to do.
Rod Elliott. What a Star. He and I seldom agree. I don't mind at all because from his stand point he is right and I see that. That is science made simple. Rod in some eyes is like a bottle of beer. That's only because he isn't out to be esoteric. He knows just as much as those guys.
Rods FET amp class AB is one of his best. I like El Cheapo 1 and 2 very much. An American amp that might have been called Corona was very similar. El Cheapo could be made into class A.