Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 6th November 2012, 07:49 PM   #41
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
I told you a bad number earlier: 4 Vp 1 Watt THD 0.00015%
This is not accurate. I did FFT with too little time setting.
Now when I allow FFT to work for a longer time
we get the low and correct value THD 0.00008% for 1 Watt.

The dist at 20kHz is not as pretty, Struth.
But I think it can be low enough ....
The circuit is fast. -3db at 3 MHz, maybe 3.5 MHz

I can't wait until tomorrow when I will try this circuit with BJT output
Quote:
Power Supply +/-42 Volt, 8 Ohm load 4e
THD 1kHz
04 Vp, 01 Watt THD 0.00008%
08 Vp, 04 Watt THD 0.00029%
12 Vp, 09 Watt THD 0.00032%
16 Vp, 16 Watt THD 0.00028%
20 Vp, 25 Watt THD 0.00026%
24 Vp, 36 Watt THD 0.00024%
28 Vp, 49 Watt THD 0.00023%
32 Vp, 64 Watt THD 0.00023%
34 Vp, 72 Watt THD 0.00024%
35 Vp, 76 Watt THD 0.00027%
36 Vp, 81 Watt THD 0.04918%

THD 20kHz
04 Vp, 01 Watt THD 0.00077%
08 Vp, 04 Watt THD 0.00432%
12 Vp, 09 Watt THD 0.00435%
16 Vp, 16 Watt THD 0.00515%
20 Vp, 25 Watt THD 0.00870%
24 Vp, 36 Watt THD 0.01649%
28 Vp, 49 Watt THD 0.03509%
32 Vp, 64 Watt THD 0.07923%
34 Vp, 72 Watt THD 0.21328%
35 Vp, 76 Watt THD 0.38580%
36 Vp, 81 Watt THD 0.66904%

Low Power 1kHz
0.5 Vp THD 0.00006%
1.0 Vp THD 0.00006%
1.5 Vp THD 0.00006%
2.0 Vp THD 0.00006%
3.0 Vp THD 0.00007%
4.0 Vp THD 0.00008% =1 Watt
5.0 Vp THD 0.00010%
6.0 Vp THD 0.00015% =2.25 Watt
7.0 Vp THD 0.00024%
8.0 Vp THD 0.00029% =4 Watt
12 Vp THD 0.00032% =9 Watt
16 Vp THD 0.00028% =16 Watt
20 Vp THD 0.00026% =25 Watt
24 Vp THD 0.00024% =36 Watt
28 Vp THD 0.00023% =49 Watt
__________________
lineup
  Reply With Quote
Old 6th November 2012, 08:10 PM   #42
Struth is offline Struth  Canada
diyAudio Member
 
Join Date: Aug 2012
Location: books at londonpower.com
Hi Guys

That's very impressive!

The BJTs I use are the MJL3281A/1302A outputs and 2SC4793/2SA1837 drivers - just in case you have similar models in your simulator.... fingers crossed...

I like the idea of the BJTs as their 'lost' supply voltage is a bit less given the lower drive voltage requirement compared to a mosfet. Also, the Vbe threshold is much more predictable than Vgs. I think z-out will be lower with BJTs.

The cost of the MJLs and IRFPs is about the same these days, so cost of one type of output stage versus another does not seem to be an issue, unless you go with 2SK/2SJ outputs that are about three to four times the cost.

Self showed a slightly lower THD with the T-triple output than with the CFP for low levels. That sways me towards the T-type EF output despite the thermal stability of the CFP.

What would really be interesting is the Bryston Quad-complimentary output stage in your buffer format. If you are unfamiliar with this, as a follower output it would be drawn as (for the upper half) an NPN driver with equal-value emitter and collector resistors (typically 33-51R). From the emitter, the base of an NPN output is driven, E to output and C to supply. From the driver collector, a PNP base is driven with C to output and E to supply. This is kind of like a Darlington overlaid with a CFP. The lower circuit half is complimentary to that described.

To compliment the buffer, we need a variable gain discrete gain stage of similar performance. No pressure.

My ideal system would have superlative performance and low noise for listening around 1W and less. A variable gain voltage stage driving a buffer seems like a good way to optimise signal to noise. Why have fixed gain interspersed with attenuations? A buffer that is either pure class-A or has a wide class-A region power and powered from +/-24V or less would suffice.

It would seem possible to overlay a current-monitoring bias servo on your buffer. Either a single Vbe or two across the output device source/emitter Rs with a BJT across the drive lines. Two BJTs in total, either way just holding Vq constant.

Have fun
Kevin O'Connor
londonpower.com

Last edited by Struth; 6th November 2012 at 08:16 PM.
  Reply With Quote
Old 6th November 2012, 10:50 PM   #43
diyAudio Member
 
Join Date: Feb 2011
Quote:
Originally Posted by MiiB View Post
in one of "thinkhlin" circuits there was som kind of crosscoupling.. Think that with some work could form some kind of error correction... have you explored that..??
Hi! MiiB, thanks for you reply!
The circuits is No problem for simulation,The solid state has not been completed,I don't how to drow PCB
  Reply With Quote
Old 7th November 2012, 03:20 AM   #44
Struth is offline Struth  Canada
diyAudio Member
 
Join Date: Aug 2012
Location: books at londonpower.com
Hi Guys

Thinkhlin, you should check out Eagle by Cadsoft. They have their version 5 PCB layout software out for a while, up to some level of modification.

I use an older version and lay out lots of boards. It was awkward to learn especially since there was a lot of basic computer stuff I didn't know at the time, but eventually I got there. All these guys with their simulation software will have no problem learning something like Eagle.

Eagle comes with various component libraries, and there are ones users have shared on Cadsoft's site. I made my own parts library as I lay out a lot of tube stuff which do not exist in the stock libraries. Eagle was originally designed around 1984 by computer guys wanting to lay out computer boards. So, there are a few terms used that come from that discipline that might throw off an analog guy at first, but you get used to the idiosyncrasies.

Even if you use only existing parts you will find it convenient to group things you use a lot into a library with a name that puts it at the top of the library list.

Hey Lineup, how about "diamond feedback buffer" for a name, if you do not want your own name on it? This makes sense since a diamond buffer is used as a feedback loop within the buffer. It also answers to the query about adding error correction, as the diamond buffer _is_ correcting the output error.

Have fun
Kevin O'Connor
londonpower.com

Last edited by Struth; 7th November 2012 at 03:22 AM.
  Reply With Quote
Old 7th November 2012, 10:59 AM   #45
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
Quote:
Originally Posted by Struth View Post
Hi Guys

That's very impressive!

The BJTs I use are the MJL3281A/1302A outputs and 2SC4793/2SA1837 drivers - just in case you have similar models in your simulator.... fingers crossed...

I like the idea of the BJTs as their 'lost' supply voltage is a bit less given the lower drive voltage requirement compared to a mosfet. Also, the Vbe threshold is much more predictable than Vgs. I think z-out will be lower with BJTs.
For the BJT tryout.
I am using MJL3281A/1302A which have been modeled by ANDY and posted in forum.
I also have the models of 2SC4793/2SA1837, but for my BJT output I will use BD139/BD140 simply. Those are good models which can be found at Home :: NXP Semiconductors
The circuit will use 100mA idle current and the compensation is now 12nF for capacitors.

I will make some figures with the BJT. I be back.
To those not convinced about this circuit I can only say that this is a good thing. And surprised I am that noone has thought about it before me.
To me a very good power buffer is gold worth. It means you can select your own frontend and present it to your loudspeakers with very little distortion. For example use a nice TUBE frontend voltage amplifier and put it to your loudspeaker through the power buffer.
__________________
lineup
  Reply With Quote
Old 7th November 2012, 11:54 AM   #46
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
On the contradictionary what is expected the BJT version is more sensitive regarding power supply. When ouput reach above 20 Vp there is an increase in distortion. I can not explain this. Because BJT would be less concerned about voltage supply.
This maybe calls for a cascoding of the current sources and those transistors.
__________________
lineup
  Reply With Quote
Old 7th November 2012, 01:51 PM   #47
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by lineup View Post
And surprised I am that noone has thought about it before me.
I think a number of people (including me) already have:
Click the image to open in full size.
I have even built a complete one (or attempted to). Actual results are quite promising, but I have already blown an impressive quantity of OP pairs which is a bit tiring, which is why it's on the back burner for the moment.
THD in LTSpice ?
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 7th November 2012, 02:59 PM   #48
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
Quote:
Originally Posted by Elvee View Post
I think a number of people (including me) already have:
Click the image to open in full size.
I have even built a complete one (or attempted to). Actual results are quite promising, but I have already blown an impressive quantity of OP pairs which is a bit tiring, which is why it's on the back burner for the moment.
THD in LTSpice ?
Great Elvee.
I hope you will get one working good. I see you have had the same thinking as me.
__________________
lineup
  Reply With Quote
Old 7th November 2012, 03:32 PM   #49
Struth is offline Struth  Canada
diyAudio Member
 
Join Date: Aug 2012
Location: books at londonpower.com
Hi Guys

Elvee, I think your circuit would benefit from using an EF output rather than CFP if only to have more voltage headroom for the error correction circuit to function. This seems to be combined with bias control and it strikes me as a "too clever" solution fraught with disappointment.

Most circuits that have active bias control of some sort would benefit from having a backup clamp circuit. This can be a simple string of diodes across the drive lines. Such an addition seems particularly important in the circuits discussed here that use current sources to turn on the output devices. If the bias control circuit opened, the outputs will be turned on hard and blow themselves up. With a diode clamp in place, maximum current can at least be limited, and hopefully the heat sink is sized to accommodate this fault condition so nothing actually blows up.

The diamond feedback path in Lineup's circuit provides error correction. I believe it to be a unique and new circuit.

The Elvee circuit is incomplete as presented. As a simulated circuit it uses ideal components that are perfectly matched. Emitter Rs will be needed in all positions, to accommodate real-world components. There seems to be no easy way to adjust bias.

To me, it seems like a flaw in most CFB designs that the idle condition in every stage is dependent on that of the input stage. This allows no optimisation for individual stages.

I never use the TO-126 case as it needs too much mounting hardware to go on a heat sink compared to the 2SC4793/2SA1837 which are F-paks. These are the same as the older 2SC3298/2SA1306 that is out of production.

Cordell intrigued everyone with his application of error correction, but the circuit is tweaky. His explanation of how to calibrate it is a bit vague but he does indicate that the benefit of the circuit drops off with frequency, just as all feedback benefits do. Unlike that performance, Lineup's diamond feedback has much less limitation, exhibited by the response to 3.5MHz noted in a previous post.

Have fun
Kevin O'Connor
londonpower.com

Last edited by Struth; 7th November 2012 at 03:37 PM.
  Reply With Quote
Old 7th November 2012, 05:51 PM   #50
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by lineup View Post
Great Elvee.
I hope you will get one working good.
Having one robust and reliable enough to be shared as a project on the forum may take some time....

Quote:
Originally Posted by Struth View Post
Hi Guys

Elvee, I think your circuit would benefit from using an EF output rather than CFP if only to have more voltage headroom for the error correction circuit to function. This seems to be combined with bias control and it strikes me as a "too clever" solution fraught with disappointment.
It doesn't really matter: all transistors work with a Vcb>=0. The circuit will absorb transparently anything you care to throw at it: MOS, darlingtons, CFPs: I have even used a quasi.

Quote:
Most circuits that have active bias control of some sort would benefit from having a backup clamp circuit. This can be a simple string of diodes across the drive lines
Yes, that is exactly what I used at the first stages as an idiot-proof safety net. But it doesn't help for severe oscillation problems that cause cross-conduction, and it has to be removed for actual performance tests because even a small parasitic current degrades the performance: the bases need to be completely independent.
Quote:
. Such an addition seems particularly important in the circuits discussed here that use current sources to turn on the output devices. If the bias control circuit opened, the outputs will be turned on hard and blow themselves up.
I know.....



Quote:
The Elvee circuit is incomplete as presented. As a simulated circuit it uses ideal components that are perfectly matched. Emitter Rs will be needed in all positions, to accommodate real-world components. There seems to be no easy way to adjust bias.
On the contrary, it is quite straightforward: I4 is a little larger than the difference between I1 and I3. The small excess sets the quiescent current: Iq=(26*ln((I1-I3)/(I4-I1+I3)))/R1.
This can be done by making I4 (and I6 obviously) adjustable

Quote:
To me, it seems like a flaw in most CFB designs that the idle condition in every stage is dependent on that of the input stage. This allows no optimisation for individual stages.
It doesn't really matter here

Quote:
Cordell intrigued everyone with his application of error correction, but the circuit is tweaky. His explanation of how to calibrate it is a bit vague but he does indicate that the benefit of the circuit drops off with frequency, just as all feedback benefits do. Unlike that performance, Lineup's diamond feedback has much less limitation, exhibited by the response to 3.5MHz noted in a previous post.
The small signal bandwidth of this type of circuit is substantial, even with the crap components I love to use. Power bandwidth is obviously more limited.
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
four transistor emitter follower (diamond buffer) as power output stage? capslock Solid State 113 12th August 2013 06:16 PM
mosfet source follower output stage. Printer2 Tubes / Valves 21 18th January 2013 04:43 AM
LINEUP Input-VAS, idea lineup Solid State 57 14th March 2011 11:33 PM
Do we always need an emitter follower to drive a LATFET output stage??? Leolabs Solid State 37 9th January 2011 04:16 PM
Cathode follower Class A output stage? Circlotron Tubes / Valves 14 15th May 2003 11:11 PM


New To Site? Need Help?

All times are GMT. The time now is 02:30 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2