The Ultimate Transistor Amplifier

[Fish4Fun]

****WOW Is that an analog scope with a CRT screen? WOW****

LMFAO.....sorry, I couldn't help myself. I will be very frank: From your initial post in this thread, I honestly dramatically underestimated you. My apologies. I applaud your efforts! As previously stated, back in the late 70's Audio Amplifiers were my inspiration into electronics. It has been quite some time since I considered designing/building one. In the late 1990's I briefly explored class D amplifiers, but once again, I was too far behind the learning curve to do anything more than take intellectual notes. I hope that you will continue to post your progress on this project; I think there are several of us quite interested (well, at the very least me and Chris).

Fish

 

[CDRIVE]

Fish, I don't have digital scope. My current scope inventory includes ..Fairchild 767H, HP 1741A and a Tektronix 221A. Over the last year or so I gave away two cart mounted vintage Tektronix models but I don't remember the models. Somewhere buried in my junk is an 80's vintage Leader with a fried flyback. My first scope in 1967 was an old black Dumont! I think it had a -3dB BW of 500KHz and manufactured in the late 40's or early 50's!

Roger, I spiced your circuit up to the Drains of the JFET drivers. With 100mVPP @ 500Hz input the THD = 4.6%. Since I was never what would be considered an audiophile I wouldn't know if this is good or bad.

I've been unable to spice the PA outputs though because I don't have a spice model for the 2SK175. It appears to be a power MOSFET switching transistor. Don't they have an almost non existent linear transfer curve?

Chris

 

[Tamal Tomal]

Thanks a lot. Its awesome. I was searching this type circuit for a long while, thanks again.

 

[rogerk8]

****WOW Is that an analog scope with a CRT screen? WOW****

LMFAO.....sorry, I couldn't help myself. I will be very frank: From your initial post in this thread, I honestly dramatically underestimated you. My apologies. I applaud your efforts! As previously stated, back in the late 70's Audio Amplifiers were my inspiration into electronics. It has been quite some time since I considered designing/building one. In the late 1990's I briefly explored class D amplifiers, but once again, I was too far behind the learning curve to do anything more than take intellectual notes. I hope that you will continue to post your progress on this project; I think there are several of us quite interested (well, at the very least me and Chris).

Fish

Thank you Fish for your kind words!

I had actually to google LMFAO

Well I am not that skilled in electronics even though I work in the area. My dream have for long been to become a HW designer and it is actually kind of happening now.

This schematic of mine has risen from my knowledge in (linear/audio) tube amplifier design and especially the Williamson amplifier. I have just applied my knowledge to modern components. Just for fun

Best regards, Roger
PS
I actually do not see the need for digital scopes whatsoever. Especially when it comes to audio frequencies. But even otherwise. I have repaired CPU-based controller cards for some ten years now. An estimation of the number of times I actually have had the need for a digital scope is less than 1%!

But of course they are nice and good att saving interesting screen-dumps. Yet there is an invention called camera

 

[rogerk8]

Fish, I don't have digital scope. My current scope inventory includes ..Fairchild 767H, HP 1741A and a Tektronix 221A. Over the last year or so I gave away two cart mounted vintage Tektronix models but I don't remember the models. Somewhere buried in my junk is an 80's vintage Leader with a fried flyback. My first scope in 1967 was an old black Dumont! I think it had a -3dB BW of 500KHz and manufactured in the late 40's or early 50's!

Awesome park of oscilloscopes! Especially your Dumont!

As I said to Fish, digital oscilloscopes are not neccesary. You can accomplish so very much just by probing a signal and recognizing that it changes.

I especially find two features of a digital scope awesome. These are Single-shot and thereby being able to capture and zoom/analyze a difficult signal and the screen-dump feature which gives you the possible to present data about a signal in an undisputable way.

But use of an analog scope and a simple (digital) camera is good enough because you can always tell the settings. They need not be proven all the time.

One major advantage with analog scopes is also the ease of getting started

Roger, I spiced your circuit up to the Drains of the JFET drivers. With 100mVPP @ 500Hz input the THD = 4.6%. Since I was never what would be considered an audiophile I wouldn't know if this is good or bad.

Thank you very much for this! This is real fun to hear! 4.6%. Not bad for a non-feedbacked transistor design!

A classic non-feedbacked 300B tube SE-amplifier will have some 4% distorsion. And many people, almost including me but I like 845 better, think that this type of power amplifier/distorsion is the best that humans can make!

I've been unable to spice the PA outputs though because I don't have a spice model for the 2SK175. It appears to be a power MOSFET switching transistor. Don't they have an almost non existent linear transfer curve?

Chris

The 2SK175s are old audio MOSFETs from the 80's. They seem to have a rds(on) of some 0,75 Ohm so I lose some 4Vds for the positive swing in Vgs. I have however calculated with this. But yes, they are real unlinear in this saturated area. This is why I need some feedback.

By the way, do they manufacture power MOS in depletion mode? Always wanted a couple of those. Because then the whole amp could be designed almost exactly as a Williamson amp.

Best regards, Roger

 

[(*steve*)]

I actually do not see the need for digital scopes whatsoever.

Firstly, my scope is an analog CRO.

However, there are times when it would just be super handy to have a digital scope. The two times that it normally makes me wish are:

1) when viewing a 1-time, or slowly repeating signal.
2) when I want to see what happens just before the trigger.

Another trick is that on digital scopes which allow you to capture the signal to a USB stick or whatever, you are sometimes able to get that, do some magic (FFT) and get a poor-man's spectrum analyser.

 

[CDRIVE]

Awesome park of oscilloscopes! Especially your Dumont!

By the way, do they manufacture power MOS in depletion mode? Always wanted a couple of those. Because then the whole amp could be designed almost exactly as a Williamson amp.

Best regards, Roger

I remember the Dumont had only single trace, AC input and no trigger controls what so ever. There was only a Sync pot. Locking the signal was near impossible.

Interesting that you mention depletion mode power MOSFETs. My Tina spice library only provides a generic model that has no characteristics favorable to power amps. I did find this reference to a power JFET though.. but where the hell is the link to the model, or even a part number???

http://www.eetimes.com/author.asp?section_id=1&doc_id=1283399

Chris

 

[CDRIVE]

This is an interesting read about power JFET's.

http://www.6moons.com/industryfeatures/zen8/zen8.html

Chris

Edit: Links to this were not readily apparent, so I thought I'd add this.

http://www.nextag.com/Power-Jfet/products-html?nxtg=2d850a3c0505-1099E80214E9E398

 

[rogerk8]

This is an interesting read about power JFET's.

http://www.6moons.com/industryfeatures/zen8/zen8.html

Chris

Edit: Links to this were not readily apparent, so I thought I'd add this.

http://www.nextag.com/Power-Jfet/products-html?nxtg=2d850a3c0505-1099E80214E9E398

Thank you very much, Chris!

I did not know there was such things as Power JFETs. At first I actually thought it was a joke

But this is a dream come true! Thanks!!

The LU1014D is surprisingly linear and I have of course downloaded the datasheet.

Your last link thankfully makes it possible for me to even order a couple. And at the rediculously low price of some $2,75!

I will contemplate this and probably redesign the output stage. Because I love JFETs!

Best regards, Roger
PS
Viewing your first link made me almost laugh (except for the Power JFET Amplifier page turning up as an "Error 404") when they said that they have cascoded MOSFETs to be able to handle the higher voltage needed. They obviously don't see the benefit of an OPT where output power may be determined by more or less the current only.

 

[CDRIVE]

PS
Viewing your first link made me almost laugh (except for the Power JFET Amplifier page turning up as an "Error 404") when they said that they have cascoded MOSFETs to be able to handle the higher voltage needed. They obviously don't see the benefit of an OPT where output power may be determined by more or less the current only.

Not being an audiophile I'll reserve judgment on this. An ideal transformer can be defined simply as its turns ratio, where; Voltage Transfer Ratio (N)= Vsec/Vpri. It follows that impedance and current follow the same convention. It also follows that P remains as a constant.

This all looks great on paper but transformers are not ideal. They're coupled inductors with very real frequency response (BW) curves. DC resistance, core saturation, eddy currents and loss aren't considered with an ideal transformer either. They're also costly, large and heavy.

BTW, the author did mention transformer output early in the article.
http://www.6moons.com/industryfeatures/zen8/zen8.html
See Fig-4 and attendant text.

You'll also note that the article included a simplified schematic (Fig1) of an old Yamaha JFET amplifier that used complimentary symmetry output.

All that said and to echo what Fish said about initially underestimating you, I did too and I'm not going to make that mistake again.

One thing that I can advise is finding a MOSFET symbol for your current schematics. Using JFET symbols will confuse those who are not familiar with the 2SK175 used in your output.

Chris

 

[CDRIVE]

I needed to refresh my memory so I did a quick spice to verify it. JFETs are ambidextrous devices. IE, the Source and Drain are interchangeable. This puts the schematic symbol your using for JFETs in question. The only arrow that should be present is in the Gate node.

Chris

Edit: The LU1014D incorporates a protection diode like MOSFETs. This would preclude them from being ambidextrous. There must be a better term to describe this though.

Chris

 

[rogerk8]

Not being an audiophile I'll reserve judgment on this. An ideal transformer can be defined simply as its turns ratio, where; Voltage Transfer Ratio (N)= Vsec/Vpri. It follows that impedance and current follow the same convention. It also follows that P remains as a constant.

This all looks great on paper but transformers are not ideal. They're coupled inductors with very real frequency response (BW) curves. DC resistance, core saturation, eddy currents and loss aren't considered with an ideal transformer either. They're also costly, large and heavy.

BTW, the author did mention transformer output early in the article.
http://www.6moons.com/industryfeatures/zen8/zen8.html
See Fig-4 and attendant text.

You'll also note that the article included a simplified schematic (Fig1) of an old Yamaha JFET amplifier that used complimentary symmetry output.

All that said and to echo what Fish said about initially underestimating you, I did too and I'm not going to make that mistake again.

One thing that I can advise is finding a MOSFET symbol for your current schematics. Using JFET symbols will confuse those who are not familiar with the 2SK175 used in your output.

Chris

Thans for the smiley! Otherwise I would have gotten depressed

I for sure know that there are problems with OPTs but I also see the benifits like no current semi-conductor amplifier designer seem to do. Of course, complementary output stages looks better on paper. But they do miss two obvious facts:

1) There is an extremely low chance that there will be DC at your loadspeaker.
2) PP-designs reduces distorsion by their nature (i think it is even harmonics but I do not know)

I have now redisigned the FET-amp because of wanted details in the design such as the possibility to exactly adjust the push-pull gain of the driver stage via P1 (will probably not be audible but yet...) and the possibility to calibrate the power output DC bias (P2&P3) at TP1 and TP2 (referring to B+ and the inherent DC-resistance of the OPT).

I think I can not do any better than this. This is my "Ultimate Transistor Amplifier".

I was so very excited about your Power J-FET and I now actually have a design for these LU1014D also but it is hard to tell what happens "after" Vgs=-1V. I have tried to extrapolate the behaviour after Vds=5V but it is kind of obvious that the quite linear behaviour for lesser Vgs is lost. An ideal load seems to be 0,1 Ohm@B+=5V. But how would I make an OPT that can handle 50A?

Anyway, I have done a preliminary design that I probably will not build. But it was fun designing it

Take care!

Best regards, Roger
PS
I care too much about what people says. But I have my way of thinking and designing stuff anyway.

 

[CDRIVE]

Roger, I hope my words were not interpreted as criticism. They're not intended to be discouraging in any manner. I'm a bit fascinated with adapting tube design to JFETs too! At this point in my life I'm not physically building circuits but I do like spicing stuff. Spice can be very accurate as long as real world parts are not ignored. Spicing transformer characteristics is something that I haven't spent much time with and it's been many decades since I studied them. So this is fun for me!

Assuming RL=4Ω please fill in these transformer variables for me so I can play with it.

Po=?
Ein (PP)=?

Chris

 

[rogerk8]

Roger, I hope my words were not interpreted as criticism. They're not intended to be discouraging in any manner. I'm a bit fascinated with adapting tube design to JFETs too! At this point in my life I'm not physically building circuits but I do like spicing stuff. Spice can be very accurate as long as real world parts are not ignored. Spicing transformer characteristics is something that I haven't spent much time with and it's been many decades since I studied them. So this is fun for me!

Assuming RL=4Ω please fill in these transformer variables for me so I can play with it.

Po=?
Ein (PP)=?

Chris

Hi Chris!

Thank you for your nice words. Now I'm feeling excellent again!

Po(tot)=10W
Ein(PP)=18Vpp (>12V btw...)
RL(PP)=8 Ohm (2 Ohm/FET)
L(PP)>0,2H

I was thinking of a favor you might want to do me. Would you mind buying four LU1014D for me? I just need to have a couple of those The problem I have is that I own no credit card which is an active choice of mine. Being a drunk and all

My plan is to build a curve tracer (which I already have designed) to take a closer look at this fabulous component with regard to Vds<12V and Id<6A.

Finally, I am attaching my FET amplifier where the MOS-symbols have been edited to your nice advice. I prefere using shown symbols because I think ordinary MOS-symbols are just too complicated.

Take care my friend!

Best regards, Roger
PS
A drawback with LU1014 is, while it is nicely biasable like a tube, it carries a lot of current. This current is higher than any ordinary power potentiometer can handle and this means that there is no way to neither balance the bias nor set the bias. Using my ordinary FET-amp it is possible to do both with low power potentiometers. Pre-matching the LU1410s is thus the "only" way.

Or a combination which however destroys the niceness of tube biasing.

Still, the linearity is amazing and the swing is ossom (almost 24Vpp in our case).

 

[CDRIVE]

At this point in my life I'm not physically building circuits but I do like spicing stuff. Chris

Hi Chris!
I was thinking of a favor you might want to do me. Would you mind buying four LU1014D for me?

What in that statement implies that I'd find joy in doing that? I just finished doing 3 days of donuts around my neighborhood. Why you ask? Because I ordered a bike maintenance fixture that was shipped via FedX, who will not leave an unattended package. The tracking link was useless, as it updated only once every 3 days. Consequently, I couldn't cycle more than around the block over and over and over and.......! Edit: Make that 4 days! I forgot to include the delivery day.

Here's a thought... How about family? Someone in your bloodline must have a credit card or internet credit.

Chris

 

[rogerk8]

What in that statement implies that I'd find joy in doing that? I just finished doing 3 days of donuts around my neighborhood. Why you ask? Because I ordered a bike maintenance fixture that was shipped via FedX, who will not leave an unattended package. The tracking link was useless, as it updated only once every 3 days. Consequently, I couldn't cycle more than around the block over and over and over and.......! Edit: Make that 4 days! I forgot to include the delivery day.

Here's a thought... How about family? Someone in your bloodline must have a credit card or internet credit.

Chris

Hi Chris!

I beg of your forgiveness. The thought was of course that I would send you an envelope with the amount.

I do have a friend that is buying stuff for me. But he recently told me that he prefers that it is possible to buy them in Sweden. Which they aren't.

Sorry for making you upset.

Friends?

Best regards, Roger
PS
I have no family here in Gothenburg. And I have very few friends.

 

[CDRIVE]

Roger, I don't understand some of your unit values, or rather the terms that you use. When you refer to the transformer use the terms that we'll all understand like "pri" & "sec" as in Lpri=.2H, Lsec=.3H. This tells me the primary is .2H and the secondary is .3H. If I read Npri=100 and Nsec=200 I'll read that as the transformer has a primary of 100 turns and the secondary has 200 turns.

Po(tot)=10W ' Is this RMS?
Ein(PP)=18Vpp (>12V btw...) ' What does btw mean?
RL(PP)=8 Ohm (2 Ohm/FET) ' There's no such thing as peak to peak ohms
L(PP)>0,2H ' Does this mean L(primary)?

Sorry for making you upset.

Friends?

Best regards, Roger

Making me upset??
Roger, my kid sister "upsets" me. A-holes texting while driving and nearly running over me and my bike "upsets" me. The guy that currently occupies the White House "UPSETS" me big time! You....? You don't upset me unless you're one of those three masquerading as RogerK8 the JFET guy.

Chris

EDIT: Yes, still friends.
EDIT: Having many friends is unimportant. Having a few good friends will trump many every time!

 

[rogerk8]

Roger, I don't understand some of your unit values, or rather the terms that you use. When you refer to the transformer use the terms that we'll all understand like "pri" & "sec" as in Lpri=.2H, Lsec=.3H. This tells me the primary is .2H and the secondary is .3H. If I read Npri=100 and Nsec=200 I'll read that as the transformer has a primary of 100 turns and the secondary has 200 turns.







Making me upset??
Roger, my kid sister "upsets" me. A-holes texting while driving and nearly running over me and my bike "upsets" me. The guy that currently occupies the White House "UPSETS" me big time! You....? You don't upset me unless you're one of those three masquerading as RogerK8 the JFET guy.

Chris

EDIT: Yes, still friends.
EDIT: Having many friends is unimportant. Having a few good friends will trump many every time!

Hi Chris!

Transformer data:

Npri=2X100=200
Nsec=141
Lpri>0,2H
Zsec=4Ohm (speaker impedance)
Zpri=8 Ohm (2 Ohm/FET)
Pout=10W Sine (5W/FET)
Uin=18Vpp (peak-to-peak, over the whole primary)

DC-loss and interlayer capacitance or leakage inductance are hard to tell at this point.

Best regards, Roger
PS
I interpreted your upper case PP as Push-Pull or Plate-to-Plate. You have now tought me that I should be more clear with parameters in the future.

You are so right. My few friends are the best friends in the whole world! And that is what it's all about

 

[CDRIVE]

Hi Chris!

Transformer data:

Npri=2X100=200
Nsec=141
Lpri>0,2H
Zsec=4Ohm (speaker impedance)
Zpri=8 Ohm (2 Ohm/FET)
Pout=10W Sine (5W/FET) ' We need to talk about this spec.
Uin=18Vpp (peak-to-peak, over the whole primary)

Hi Roger. When I first read this I thought you were misinterpreting or misstating this spec. I started typing up a reply but I've deleted all of that because I don't want my statements to be wrong. I'm going to mull this over a bit before I commit to an opinion. Heck, you may be correct and I'm wrong!

Cheers,
Chris

 

[rogerk8]

Hi Chris!

I have now redesigned my FET-amp somewhat. The redesign incorporates the possibillity to run the amp on both J-FETs and MOS-FETs (that is the LU1014D and the 2SK175).

This is just a prototype with a minimum of complexity for testing of them both.

The power bias calibration possibillity is probably not that important. But in view of the high currents the unair-gapped OPT might easily become saturated even though the number of turns are few. Anyway I may set the accurate bias as I please and the MOS-current might actually be measured at power supply. If not as voltage across the primary winding...

Best regards, Roger
PS
I spoke to my friend today. He now says that abroad ordering is not a problem