F5 power amplifier

Ex-Moderator R.I.P.
Joined 2005
Hmm, actually, I was first thinking about taking a 4U 400mm.

each heatsink on the 400mm design is made from two pieces, and best avoided

from what I can see the price difference between 4U and 5U in 300mm is only 30EUR
I would definately choose 5U, no doubt about it
5U is the one you want, and need

actually, prices on 4U/400mm and 5U/300mm are exactly the same, so I dont understand your argument
no question about it, heatsinks are not where you want to save money

if only the 400mm had properly designed heatsinks
 
Hi everyone,

I plan to finalyse my BoM and then purchase the last little things I need.
Nevertheless, I have 2 questions about 1) the transformer 2) the case.

  1. according to several threads on several forums I read, it seems that a 400VA instead of a 300VA would be better. Is that true?
  2. I plan to buy either a HiFi2000 4U 300mm or a 4U 400mm. According to the attached file, a 4U 300mm should be ok, but isn't it a little bit too tight?
Thank you very much in advance.

I'm using the 4U 300mm chassis and is perfect for the amp, enough dissipation for the Madrid summer.

3U is too small
 
from what I can see the price difference between 4U and 5U in 300mm is only 30EUR
I would definately choose 5U, no doubt about it 5U is the one you want, and need

actually, prices on 4U/400mm and 5U/300mm are exactly the same, so I dont understand your argument
no question about it, heatsinks are not where you want to save money

if only the 400mm had properly designed heatsinks

Hello,

first of all, my mistake, I did not use the proper word. By "tight", I wasn't talking about the dimensions of the case, but I was talking about its power dissipation capacity.

Here is the website I will order the case. I can see :

  • 4U 300mm = 119,50€ (0.31 °/W and 165W Max total power dissipation for 2xTO3-P transistors on each sink)
  • 4U 400mm = 154,00€ (0.23 °/W and 175W Max total power dissipation for 2xTO3-P transistors on each sink)
  • I don't see any 5U 300mm on that shop, and the cheapest 5U is 200€.
So, according to this shop, the difference between those cases is not unimportant, specially about a 5U.
Moreover, it seems that the 4U 300mm fills all the required conditions. All I wanted is the be sure I can chose it :)

Anyway, again, my mistake.

I'm using the 4U 300mm chassis and is perfect for the amp, enough dissipation for the Madrid summer.

Good to hear. :)
Thx a lot.
 
transformer VA

Normally a ClassA amplifier has a transformer rated ~ 6 to 10 times the maximum ClassA output power.
300VA would provide adequate power for 30W to 50W of total ClassA power.
400VA would provide adequate power for 40W to 65W of total ClassA power.

There is an alternative method of assessing the required transformer rating.
300VA 18+18Vac gives a current rating of 8.33Aac. This reduces to ~4.17Adc after the capacitor input filter. The 400VA would be rated at 5.55Adc.
The 1.3Adc of a single channel of F5 would use ~31% of the 300VA transformer capability. and use ~23% of the 400VA transformer capability.

A 400VA should be able to power two channels of F5, whereas the 300VA is more than enough for a single channel of F5.

Both assessments show that there is no technical reason for using bigger transformers.

There is the possibility/probability that sound quality might change, but is 33% bigger going to show that effect when both sizes are already bigger than needed for a single F5.

If you were to compare 200VA to 400VA powering a single channel F5, I would expect a difference, even though technically both are big enough.
 

This link I don't believe answers the question ;) It only talks about getting better complementary behaviour from the input stage.
The major difference in setting up the output for Toshiba's is in the Vgs. The IR FETs and most of the other mfg's we typically use have to have 4Vgs or so to get turned on to 1.3A + or -. The Toshiba MOSFETs are going to want to have maybe 2V or so. This means that the drain Rs of the JFETs needs to be smaller. Or, the pot will be adjusted to a very low value swamping out the paralell fixed resister. I don't like pots. I would take it out once I found the necessary value of the Input drain R's. But, it might be best to just reduce the value of the fixed Rs to 1k or so, so more current goes through them than the pot. If you read through this thread and/or the balanced F5 Question thread, you will find JFET drain R values that have actually been used with Toshiba outputs. You would probably have found an expanation similar to this and you wouln't be asking. If you are interested in changing the output Source resistors to a lower value you will need an even smaller JFET drain R value.
 
This link I don't believe answers the question ;) It only talks about getting better complementary behaviour from the input stage.
The major difference in setting up the output for Toshiba's is in the Vgs. The IR FETs and most of the other mfg's we typically use have to have 4Vgs or so to get turned on to 1.3A + or -. The Toshiba MOSFETs are going to want to have maybe 2V or so. This means that the drain Rs of the JFETs needs to be smaller. Or, the pot will be adjusted to a very low value swamping out the paralell fixed resister. I don't like pots. I would take it out once I found the necessary value of the Input drain R's. But, it might be best to just reduce the value of the fixed Rs to 1k or so, so more current goes through them than the pot. If you read through this thread and/or the balanced F5 Question thread, you will find JFET drain R values that have actually been used with Toshiba outputs. You would probably have found an expanation similar to this and you wouln't be asking. If you are interested in changing the output Source resistors to a lower value you will need an even smaller JFET drain R value.

Whilst your explanation may be correct, I have built a F5 without any change to the schematic using a Toshiba pair based on my readings in this forum. There are enough questions and responses on the same matter from me in this forum as well.

No problems with setting up the bias with the trimpot. You do not need to implement even the 'degeneration' resistor trick to work. The trick I believe gets better linearity with the JFETs. I have implemented the 5ohms resistor with J74. Sounds wonderful. I do not have a distortion analyzer.
 
I suppose I might have suggested anivla's point. I beleive I've read that before. There is a range to most of the device parameters we have to deal with. It is possible you might not be close enough with certain devices to operate and set bias just by swapping transisters, and not changing a few values of the passives.
I however would do what I suggested thinking it "should be" changed. Actually, I want to complete a Toshiba version with 2 pair in the output, smaller value source Rs and maybe 1.6A each for a total of 3.2A Iq with 19V rails.
Due to the reduced gain of the input with smaller JFET Drain Rs, I would like to experiment with a pair of input JFETs (2X74s, 2X170s) of lower Idss to get the gain back up. I'll start with a stock front end first, then try the 5ohm mod. Then maybe try the doubled up input?
I have CViller boards stuffed with .27 ohm Rs in the outputs and a 620 ohm in parallel with a 5k Pot on the drain of the JFETs. :D
 
Ok guys, I have a moment of insanity. Here is my offer: I volunteer to document some variation of the circuit, to be published after the review of an expert member, in the most appropiate format (post, article, wiki, whatsoever). The goal is to relief the expert member from recurring questions and give info without the need of reading 10000 posts.

Davide
 
I suppose I might have suggested anivla's point. I beleive I've read that before. There is a range to most of the device parameters we have to deal with. It is possible you might not be close enough with certain devices to operate and set bias just by swapping transisters, and not changing a few values of the passives.
I however would do what I suggested thinking it "should be" changed. Actually, I want to complete a Toshiba version with 2 pair in the output, smaller value source Rs and maybe 1.6A each for a total of 3.2A Iq with 19V rails.
Due to the reduced gain of the input with smaller JFET Drain Rs, I would like to experiment with a pair of input JFETs (2X74s, 2X170s) of lower Idss to get the gain back up. I'll start with a stock front end first, then try the 5ohm mod. Then maybe try the doubled up input?
I have CViller boards stuffed with .27 ohm Rs in the outputs and a 620 ohm in parallel with a 5k Pot on the drain of the JFETs. :D

Will track your efforts. I am in the process of doing a paralleled Toshiba SE F5 as well. Talking of 0.22ohms source resistors in the outputs for the Toshiba, I have seen a post from Mr. Pass and I quote "Judging by the curves on those parts, when paralleling them you can raise the Source resistor to 1 ohm and get greater thermal stability yet." This suggestion is going in the other direction from 0.22 ohms. What to make of it? I am sure the Master is right.

Cheers.
 
The F5 is DC coupled.
You have effectively removed two high pass filters from your system. You are now hearing the difference between a very slightly rolled off bass with a medium rolled off bass. And this with the same speakers that cannot reproduce those very low bass frequencies.

Bah.

No way imo.

The difference is in Damping Factor!!



_-_-bear
 
anilva, I think stability would improve but there are other factors to think about. I think though the difference will be Aol. Euvl's measurements show slightly less gain(Transconductance) with the Toshiba's. But, 2 of them certainly corrects that somewhat, so you can afford slightly larger Rs. And, although loosing some gain in the input stage with smaller Drain Rs the output stage gain with a partallel pair is correcting for the input stage lower gain. I think maybe more correction in input gain may still be necessary. I would like to see a little higher closed loop gain though, without throwing it all away in feedback. About 10(20db) would good :D
 
Last edited:
Tanks Anilva Flg Nikon (in alphabetical order)
I am new to the 3 leged kind and realy apreciate all your efforts.

From the last few post You mention that Jfets with lower IDS give more gain
Could you please esplain a bit more about this.

I am working on a 28V rails F5 at present with the Farchild and waiting for the Tosh to clear custom.

Tanks
 
Member
Joined 2009
Paid Member
Ok guys, I have a moment of insanity. Here is my offer: I volunteer to document some variation of the circuit, to be published after the review of an expert member, in the most appropiate format (post, article, wiki, whatsoever). The goal is to relief the expert member from recurring questions and give info without the need of reading 10000 posts.

Davide

I for one would appreciate something like this David.
 
...From the last few post You mention that Jfets with lower IDS give more gain
Could you please esplain a bit more about this...

There is very little variation in gain between Low and high Idss JFETs. What I meant was to parallel 2 JFETs to increase gain. If I have 2 ll JFETs the input stage current will double so I would have to use lower Idss devices to keep the Idss somewhere close to the stock value. If not, I would need an even lower value Drain R loosing more gain. There could be another problem with ll JFETs, and that would be, how to connect the source feedback arangement?