JLH 10 Watt class A amplifier

Hello everyone, listening test of my DIY amplifier (JLH 1969 10W monoblock)
2x monoblock (stereo)

speakers 6 ohms
I Q. bias 2A
Power supply: 18V DC (6A)
1kHz = 524mV (-3.5dB) (Maximum input without distortion)
1kHz = 6.45V (output distortion free limit)
Pure CLASS A
Power 7W RMS (per channel)

JLH 1969 - TESTE 9 - Audição - Amplificador "CLASSE A" PURO • (10W) • MONOBLOCO

DIAGRAM (CIRCUIT PDF)
https://drive.google.com/drive/folders/1hqNqf1LpdyNb26nMPSwo5KOk3z-7M26V
 
Trust me you don't need any kit or pcbs, if you know how to solder then you can assemble it on veroboard. For TR4 you can use almost any small signal high gain pnp transistor like BC××× series or plastic package 2N3906, for TR3 BD139-16 or similar and for TR1-TR2 TIP3055, 2N3055, MJ15003 etc etc. Btw use 220k preset for R5, don't forget that.

Screenshot_20220724-140759~3.png
Screenshot_20220819-033237~2.png
 
Last edited:
  • Like
Reactions: 2 users
Fig1.
I found out that I can move the feedback point to point A or point B and the amp still works fine.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
Can someone help me analyze if such modification makes sense? I just know it increases the input impedance.
Fig1-JLH2000_FB.png


Fig2.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.
The idea came from the ESP website: https://sound-au.com/project217.htm
Can someone test it on your circuit and share how it sounds?

Fig2-JLH2000_DUO-BETA.png
 
  • Like
Reactions: 1 user
Fig1.
I found out that I can move the feedback point to point A or point B and the amp still works fine.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
Can someone help me analyze if such modification makes sense? I just know it increases the input impedance.
View attachment 1092857

Fig2.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.
The idea came from the ESP website: https://sound-au.com/project217.htm
Can someone test it on your circuit and share how it sounds?

View attachment 1092858
You could remove C5. It WILL sound much cleaner.
Connect emitter to base of Q2 ("quasi" Darlington). It will sound more fat, round, colored, powerful, but less detailed.
Feedback Fig. 2 WILL sound less detailed, disorganized.
 
I found out that I can move the feedback point to point A or point B and the amp still works fine.
yes, it will work, just add to your story that for the amplifier to work with this inclusion, you need to bring the RV1 trimmer almost to 0, so that half of the supply voltage is obtained at the Q2 emitter.
The current of the output transistors will then increase to 4 amperes. You can reduce the current by increasing the resistance R4, but what do we get in the end?
the circuit becomes single-ended for the main input transistor Q1, and Q2 becomes the load element for Q1.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
this is very strange, because in the case of such a modification, the level of distortion increases by a factor of 10, but the spectrum itself becomes very short with the dominance of the second harmonic.
if such modification makes sense?
my opinion - doesn't make sense, Nelson Pass version of single ended circuits would sound much better.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.

about the best low-frequency response, I would argue, because. in this embodiment, in addition to an even greater decrease in the input current for Q4, the ratio R6 / R9 decreases, and to a large extent weakens the effect of the general feedback. That is, with the added resistor R91, the depth of the total NFB will be only 5 dB.
share how it sounds?

how all this sounds can be indirectly determined by the transient response, which in none of the presented options speaks of a qualitative improvement in characteristics that can be transcribed to the finished product. firstly - a completely open input for the entire frequency spectrum, which will overload the input transistor for any reason, and secondly, the speed in the feedback loop is not very high, it does not keep up with the signal ...

Can this be fixed? - of course, this is done very simply, an integration circuit at the input, an accelerating circuit from the output - and that's it .... the standard vintage circuit of John Linsey Good will again have a high-quality transient response and, as a result, good sound ....
 
Fig1.
I found out that I can move the feedback point to point A or point B and the amp still works fine.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
Can someone help me analyze if such modification makes sense? I just know it increases the input impedance.
View attachment 1092857

Fig2.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.
The idea came from the ESP website: https://sound-au.com/project217.htm
Can someone test it on your circuit and share how it sounds?

View attachment 1092858
... and do not use any transistors in PO-3 package;-) These do sound most terrible;-)
 
Fig1.
I found out that I can move the feedback point to point A or point B and the amp still works fine.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
Can someone help me analyze if such modification makes sense? I just know it increases the input impedance.
View attachment 1092857

Fig2.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.
The idea came from the ESP website: https://sound-au.com/project217.htm
Can someone test it on your circuit and share how it sounds?

View attachment 1092858
Why?
Your fig 1 alters the DC currents significantly and will need some considerable adjustment to maintain a DC level.
Point A may not make much difference as C5 couples the output through to the feedback.
Point B is somewhat senseless because you are no longer correcting the output signal, and will be attempting to control the base of Q2 which means that its emitter voltage is outside the feedback loop. JLH's circuit relies on Q2 and Q1 operating in push-pull to cancel some of the distortions.
As regards the split feedback approach, the higher resistance of R9 also needs some considerable adjustment to the bias to maintain the output at mid-rail.
There is a downside to taking the feedback from the output side of the output capacitor. Yes, it will extend the low frequency bandwidth but at the cost of lmiiting the available output power at those lower freuqencies. THe feedback tries to compensate for RC droop, so the voltage across the capacitor will change, and that will limit the output power. So your apparent improvement in frequency response comes at a cost of lower possible output power.
However, the RC droop will occur anyway, but the feedback will exacerbate it.
A better solution is to use a bigger capacitor.
I suspect that modern computer grade low ESL/ESR types will not contribute significant distortion which would warrant taking feedback from the output side.
 
  • Like
Reactions: 1 user
Fig1.
I found out that I can move the feedback point to point A or point B and the amp still works fine.
I tested it on my amp circuit, but I didn't hear any difference from the original circuit.
Can someone help me analyze if such modification makes sense? I just know it increases the input impedance.
View attachment 1092857

Fig2.
As far as I know the duo-beta feedback network is from Luxman,
I haven't actually tested this circuit, but the simulation shows it has better low frequency response.
The idea came from the ESP website: https://sound-au.com/project217.htm
Can someone test it on your circuit and share how it sounds?

View attachment 1092858
I would remove (R1), R2, R3, R5, R6, R7, R9, RV2, C1, C2, C5, Q4.
I would set RV1: output - base Q3 - (ground): does not set "Iq."!
I would use three times the same transistors: TO-220 types or Fullpak or TO-126.
Never bridge big caps with little ones.
I would reduce Vcc: 15 V or so.
And: NEVER use pcbs or veroboards.
;-)
Do use fullrange speakers;-) Without any crossover! Without any electrical frequency correction!
;-)
 
Last edited:
I think it's worth posting my simulated circuit showing that these modifications work.
The Multisim simulation files are placed in the attachment.

In mod0, I changed the parameters of JLH1969.
mod0-thd.png

mod0-fp.png



In mod1, I changed the feedback point to point A.
mod1-thd.png



In mod2, I added multiple feedback.
mod2-thd.png

mod2-fp.png
 

Attachments

  • JLH1969-mod2.zip
    156.3 KB · Views: 14
Last edited:
Everything should be made as simple as possible in other words, completely agree.

But the quote continues but no simpler

There are many compelling reasons to use a well-designed PCB and no evidence that it's going to damage the signal, so not using one for ideological reasons is ... well, eccentric.

The exception being simple circuits with very few components, where it's just not worth the (small) extra cost.
 
Everything should be made as simple as possible in other words, completely agree.

But the quote continues but no simpler

There are many compelling reasons to use a well-designed PCB and no evidence that it's going to damage the signal, so not using one for ideological reasons is ... well, eccentric.

The exception being simple circuits with very few components, where it's just not worth the (small) extra cost.
;-)
The "evidence that it's going to damage the signal": listen music. The more of paths, solder points, "corners and edges", resonances of materials... we do listen materials first.
If a setup needs so many components that a pcb becomes necessary, it is too complex to amplify a music signal cleanly;-)
 
I won't comment on your second paragraph, but I know of no evidence that a properly designed PCB can damage the 'sound' of an amplifier, unless we believe that electricity somehow behaves differently if the signal being processed is audio - and if so, how does the electricity know that there's a microphone, a loudspeaker and an ear outside of the board?
 
  • Like
Reactions: 1 user