F4 power amplifier

[cal3713]

Here are some pics. The stack is the LL7905 transformer, the input buffer, the second buffer (the one after the transformer), and the biasing section of the OPS. The picture shows an LSK389 input section. It is not the input section I am currently using (which is shown in the schematic), but it does show the PtP wiring. The other boards in the stack are of similar component count and complexity.

Thanks for posting. I love the stacks...

 

[cal3713]

First measure voltage drop across 10k resistor R6 or R7.
This will tell you total current eg 12V/10k = 1.2mA

Now determine current in programming resistors = 2.5/(P1+R9)
Let's assume P1 + R9 = 10k
So current would 2.5/10k = 0.25mA

So cathode current would be 1.2 - 0.25 = 0.95mA

That would be bad.
So simplifying things, you want to have a 13V drop across the 10k resistors if using TL431.

Or use those devices linked earlier. They require no more than 100uA, to operate correctly.

Thanks for writing that all out. Super useful and easy to copy and test.

 

[ra7]

First measure voltage drop across 10k resistor R6 or R7.
This will tell you total current eg 12V/10k = 1.2mA

Now determine current in programming resistors = 2.5/(P1+R9)
Let's assume P1 + R9 = 10k
So current would 2.5/10k = 0.25mA

So cathode current would be 1.2 - 0.25 = 0.95mA

That would be bad.
So simplifying things, you want to have a 13V drop across the 10k resistors if using TL431.

Or use those devices linked earlier. They require no more than 100uA, to operate correctly.

Thanks Rob and Pico! That makes sense.

So, what sets the current through R6 and R7, the 10k resistors? I'm trying to calculate using the schematic but cannot figure it out. Is it the ratio of R6 and the impedance provided by the JFET that sets the current through R6?

 

[2 picoDumbs]

You have a voltage drop across R5 and R22/P2 due to current draw of Jfet + current draw of shunt reference (TL431)

Let's say jfet is drawing 5mA and shunt is drawing what is required to operate, just say it's 2mA.

Then the voltage drop is 0.007 x 1000 = 7V
If you have 23V rails you now 16V left to feed the shunt.

This is not mathematically possible based on the previous assumptions, it is just purely to illustrate the situation.

To illustrate further the 10k resistor is then used to set the current, if we needed 5V for the positive rail and 5V for the negative rail to set bias for the mosfets, then we would be dropping 16 - 5 = 11V across the 10k resistor which is 1.1mA (not 2mA - that was just to keep the math simple).

 

[ra7]

Thanks! Now I know why you were recommending the other options. Of course, the simplest option is getting the right TL431.

 

[PKI]

You have a voltage drop across R5 and R22/P2 due to current draw of Jfet + current draw of shunt reference (TL431)

Let's say jfet is drawing 5mA and shunt is drawing what is required to operate, just say it's 2mA.

Then the voltage drop is 0.007 x 1000 = 7V
If you have 23V rails you now 16V left to feed the shunt.

This is not mathematically possible based on the previous assumptions, it is just purely to illustrate the situation.

To illustrate further the 10k resistor is then used to set the current, if we needed 5V for the positive rail and 5V for the negative rail to set bias for the mosfets, then we would be dropping 16 - 5 = 11V across the 10k resistor which is 1.1mA (not 2mA - that was just to keep the math simple).

So I got my mouser package with these:
1. TLVH431CLPR: https://www.ti.com/lit/ds/symlink/t...ll-mousermode-df-pf-null-wwe&ts=1601576224307
2. NCP431ACLPRAG: https://www.mouser.com/datasheet/2/308/NCP431_D-1387799.pdf
3. AP431SAZTR-G1: https://www.mouser.com/datasheet/2/115/AP431S-1114834.pdf
Which one I use?

Thanks for your help BTW, really appreciate it!

 

[Dennis Hui]

Your first part has a low reference voltage (1.25V vs the usual 2.5V)
so you'll need some adjustments to the programming resistors for biasing.

I would stick to 2 or 3. Either one looks fine to me on a quick glance.

 

[ra7]

#2 has lowest current need at 40 uA. That or #3 would work.

 

[2 picoDumbs]

So I got my mouser package with these:
1. TLVH431CLPR: https://www.ti.com/lit/ds/symlink/t...ll-mousermode-df-pf-null-wwe&ts=1601576224307
2. NCP431ACLPRAG: https://www.mouser.com/datasheet/2/308/NCP431_D-1387799.pdf
3. AP431SAZTR-G1: https://www.mouser.com/datasheet/2/115/AP431S-1114834.pdf
Which one I use?

Thanks for your help BTW, really appreciate it!

May as well start with 2

TLVH has a 1.25V volt reference, in order to use that you would also need to double the value of R8.

So the simplest to install will be 2 or 3.

 

[PKI]

Your first part has a low reference voltage (1.25V vs the usual 2.5V)
so you'll need some adjustments to the programming resistors for biasing.

I would stick to 2 or 3. Either one looks fine to me on a quick glance.

#2 has lowest current need at 40 uA. That or #3 would work.

May as well start with 2

TLVH has a 1.25V volt reference, in order to use that you would also need to double the value of R8.

So the simplest to install will be 2 or 3.

thank you so much!

 

[cal3713]

I needed an extra resistor two days ago and decided to add #2 above to my order just in case. Glad to see them recommended.

Got my parts today, measured the amp and indeed I was under the 1mA limit of the standard tl431. Curious how replacing will change the sound of the amp.

If so many people are struggling to get the tl431 to work appropriately, seems these parts should really just be part of the regular bom and installed as standard practice.

 

[2 picoDumbs]

I agree.

 

[ra7]

Also in Jim’s guide, if you’re listening Jim. That and also the JFET grade should be mentioned.

 

[cal3713]

I agree.

Thank you so much for writing about this issue. The amp sounds much better. And thanks for timing Ra, really glad you initiated the conversation.

 

[2 picoDumbs]

I am very happy to help out.
Awesome work.

 

[ra7]

Yes, a team effort. Thanks to Pico for raising the issue of the correct TL431.

I plugged the correct one in yesterday and measured 2.5V. Listened to it, didn’t notice a huge change from before. It sounds lovely, if a little hot on the top end compared to the M2. I think the M2 has a slight roll off above 10 kHz. It is also revealing that the RME is clean but a little clinical. Cleaning up the layout further will help. I am also building a 26 pre and probably the Luminaria after that, which should be fun.

I have it on the bench today. Will be rebiasing it and will report back.

 

[2 picoDumbs]

Cool.

1) Do another FFT.
Want to see how much progress you have made improving it.

2) Once you have it all biased, give it an hour to hit thermal equilibrium and give it a listen.

3) After that, retighten mosfet mounting bolts one more time, then do one more rebias/dc offset adjustment.

Did you order the Antek Can/Cover ?

 

[cal3713]

Yes, a team effort. Thanks to Pico for raising the issue of the correct TL431.

I plugged the correct one in yesterday and measured 2.5V. Listened to it, didn’t notice a huge change from before. It sounds lovely, if a little hot on the top end compared to the M2. I think the M2 has a slight roll off above 10 kHz. It is also revealing that the RME is clean but a little clinical. Cleaning up the layout further will help. I am also building a 26 pre and probably the Luminaria after that, which should be fun.

I have it on the bench today. Will be rebiasing it and will report back.

You know what they say. The f4 sounds like what's upstream. Other amplifiers will sound more like themselves...

 

[analog_sa]

You know what they say. The f4 sounds like what's upstream.

But what if you have no electrolytic coupling caps upstream?

 

[2 picoDumbs]

Is this like a Zen question?