M2HA

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I think his idea is that the Tx will drive the gates via C12, C17, loaded by R6 and R24, for AC signals.
And then by disconnecting C12 from R23 at the positive pole, the opamp can then drive the DC offset via the 4 resistors.
But it is isolated from the Tx at DC by C12 & C17. The Tx is low impedance to Gnd at DC.

Might also work.
But the "small cap" after R42 is additional load for the Tx in the audio band.


Patrick
 
Yes I know, I kept c2 in the circuit for the same reason. My circuit was basically the same as yours. I started over and now both ways work fine. They both have exactly the same distortion. The only difference is one doesn't have c2 in the circuit. I don't know why the circuit that looked like yours didn't work the first time.
Yes I know, a small cap after r42 in your circuit above is a bad idea. In the other circuit without c2 and that trace gone it made no difference with it in or out so I left it out.
 
Well I had thought I would only try to explain it since it was just a very small change to what you had already posted. I was a bit clumsy at explaining it. Still except for maybe Zen people will figure it out. Basically I wanted to use a dc servo but without the need for c2.
Also some of the models I have such as the jfets used in the circuit aren't match well so I don't like to post with them. Do you know how to modify models to match components?
 
Sorry boys.

I had a terrible week on my work and I didn´t nothing on the amp.

On the PSU, I removed Q1~4 and I bridged it, removed led´s and I changed LM337/317 for other too but I have the same problem, voltaje drop to 14,80volt and -14,40volt. So I guess that I have a problem with the msr860 or maybe on the 5K trimpots that we have on R2. :mad:
 

Attachments

  • 317_337 Cap Multiplier.png
    317_337 Cap Multiplier.png
    37.6 KB · Views: 244
This latest schematics you posted is not the same as in post#1 (Nazar shunt).
This is just a normal LM317/337 followed by a BJT cap multiplier, as used in the F5-HA.
The cap multiplier works in open loop and has some finite output impedance.
So some voltage drop is normal.

This has been proven to work in multiple examples of F5-HA using our PCBs.
So I cannot understand what your problem is.
If the output voltage is stable you can just trim out the DC offset as mentioned before.
Or just use the DC servo.


Patrick
 
Thanks Patrick

This latest schematics you posted is not the same as in post#1 (Nazar shunt).

I confused the schemes, the last one that I have set is the correct one.

This is just a normal LM317/337 followed by a BJT cap multiplier, as used in the F5-HA.
The cap multiplier works in open loop and has some finite output impedance.
So some voltage drop is normal.

This has been proven to work in multiple examples of F5-HA using our PCBs.

I didn´t know that the voltage drop was normal :eek:. I thought this was a problem with my PSU. I can compensate the output voltage of each rail with the trimpot but I never get to have 15 volts in the output. The maximum voltage I get is 14'80volt in the negative rail and about 15volt in the positive.

When both rails were decompensated, the offset in the amplifier was terrible

If the output voltage is stable you can just trim out the DC offset as mentioned before.
Or just use the DC servo.


Patrick

I will try to use the amplifier with B + -14,5 volt and then I will eliminate the C2.
 
You can also adjust the output voltage under load by adjusting the values of R2 and R7 in the power supply schematics.

Connect the PSU to the amp (without headphones).
Use a 5k trimmer, preset to ~2.7k in place of R2 and R7.
Adjust the trimmers until you get exactly 15.0V at the regulator output.
Then measure the values and replace with fixed resistors.

When this is done, then measure the output voltage of the amp with the input shorted to Gnd and C2 bypassed.
Adjust the trimmer as you have before.
Let it run for an hour and see if it is stable over time.

Easiest is still to use a DC servo.


Patrick
 
Thanks Patrick

You can also adjust the output voltage under load by adjusting the values of R2 and R7 in the power supply schematics.

Connect the PSU to the amp (without headphones).
Use a 5k trimmer, preset to ~2.7k in place of R2 and R7.
Adjust the trimmers until you get exactly 15.0V at the regulator output.
Then measure the values and replace with fixed resistors.

Ok, Done. 2.7K in r2 and r7. 15 Volt on regulators output. (13,6 volt both rails in the Psu output).

Why measure on the regulator output and not on the PSU output? And why Fer has 15volt on the output if he uses the same PSU?:confused:

I understand that my problem has been trying to reach 15 volts on the PSU's output. For some reason, I didn´t get it on the negative rail and it caused a mismatch in the offset.

When this is done, then measure the output voltage of the amp with the input shorted to Gnd and C2 bypassed.
Adjust the trimmer as you have before.
Let it run for an hour and see if it is stable over time.

Easiest is still to use a DC servo.


Patrick

Done. Now offset is very very stable. Around 0~2mVolt all time with a soft drift (I used all parts matched) :eek::eek::eek:

I need to check with headphones but now it looks good.:xfingers:
 
> Why measure on the regulator output and not on the PSU output?

I meant PSU output, sorry for the misunderstanding.

> And why Fer has 15volt on the output if he uses the same PSU?

I have no idea.
Did you measure (match) the hfe of the transistors in the regulators ?


Patrick
 
Last edited:
I can´t find the issue of this PSU.

I use 2SC4793 (Q2), BC550 (Q1) and 2SA1837 (Q4), BC560 (Q3) like F5ha PSU too.

I need use arround 17 volt on the output regulator for measure 14volt on the out put and trafo work a bit hot (15Vac@2Amp) It´s not normal.

I think I'm going to forget about these plates and use a simpler PSU like the cheap PSU I used to test the amps.:redhot::redhot::redhot:
 
> I need use arround 17 volt on the output regulator for measure 14volt on the output
That is certainly not normal.

> and trafo work a bit hot (15Vac@2Amp) It´s not normal.
Why would it draw 2A when the bias is only 160mA ?
Also not normal.

But as I can tell they are not my PCBs, so not much I can do for you.
The regulator circuit is certainly not the cause of your problems.

> I think I'm going to forget about these plates and use a simpler PSU.
Both followers have good PSRR.
So I would just use a standard 317/337 for now if I were you.

And it is always wise to test the PSUs on their own before putting them to an amp.
You can check all the DC voltages to start with.

Your previous post :
" 15 Volt on regulators output. (13,6 volt both rails in the Psu output)."
is normal. I expect a difference of ~1.3V between regulator out and PSU out.


Patrick
 
Lateral FET Output Follower

The basic M2 topology is essentially a JFET input buffer, followed by an audio transformer, and then a MOSFET power follower to drive the load.

As already mentioned in post #26, the loading on the input JFETs is quite heavy because they see R6 and R7 in parallel via the transformer. To alleviate this, one can replace R6, R7 through two constant current sources, which have higher dynamic impedance. But in order to avoid unbalanced current leading to DC offset, they need to be equal and to track each other at all times. In the Pioneer SL headphone amplifier, we have proven the stability of a current-mirror based, self-tracking CCS pair. This would be ideal to replace R6 and R7. If one is not worried about voltage headroom, once can improve this even further by using a 4-transistor full Wilson current mirror.

The capacitance of the output MOSFETs appears as additional dynamic load to the JFETs. Thus it is always advantageous to use devices with low (and constant) capacitances to reduce distortion. In that respect, lateral MOSFETs are superior. In the UTHAiM, as well as a recent (not yet published) shootout between the Borbely EB602/200 and a vertical MOSFET equivalent, multiple testers have observed noticeably smoother mids and highs. (No real surprise as Charles Hansen had been telling us all the time.) Also, as we have already discussed at the UTHAiM thread, the lateral MOSFETs have negative tempco at over 100mA bias, and is therefore self-stabilising. This will allow simply using resistors for bias, as in the UTHAiM.

So for me personally, the circuit below would be a good alternative to the output stage of post #26. But then you should not call this M2 anymore. :)

The 2SK1058/2SJ162 are already NRND, though not yet quite EOL. But the Exicon equivalents are still available and arguably better.


Patrick
 

Attachments

  • Lateral Follower V1 THD.asc
    5.3 KB · Views: 99
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.