Amplifier based on 2N3055 - Page 37 - diyAudio
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 20th January 2014, 11:26 AM   #361
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Quote:
Originally Posted by AndrewT View Post
They are confirming they are not complementary.

That is why you will see a technical argument in favour of quasi complementary, where you design the quasi to perform as you want, using devices that are near enough identical.
That's what I was afraid of. Building an amp with those MJ15015/16 as complementary parts is thus not so practical and would not result in anything really linear.

We're ok here on this quasi build, even if someone opts to make use of the MJ15015 instead of the 2N3055, so the rails can be pumped up a bunch.

Perhaps the MJ2955 is a much closer match to the 2N3055. But then there is no way to raise the rails.

I really think a good reliable tester that can properly and non-destructively test for second breakdown, along with other parameter matching, is warranted, so parts can not only be matched, but also be carefully selected for their upper limits and used safely with chosen power rails. We all know there are many parts that will withstand far more than the very limited 60V Vce0 in the specs, but testing must be done to make sure.
  Reply With Quote
Old 20th January 2014, 11:32 PM   #362
diyAudio Member
 
Join Date: Sep 2006
Hi
ON semi (Motorola) used to put the 2N3055 with the MJ15015 on a data sheet. But the recent 2N3055/MJ2955 data sheet gives both having an fT of 2.5MHz (min).

I suspect that the older MJ15015 devices were intended to match the second breakdown performance of the old, original RCA 2N3055 (later 2N3055H). This means that they probably did use a wider base with lower fT. This is probably why some people still think the 2N3055 is "not good" for hifi.

The epi base devices which have fT's around 3Mhz are closer. BUt rather than using MJ15015/6's for higher voltages (which may be slow) I'd suggest using MJ15003/4 or going straight to the MJ21193/4 which have 4MHz fT's and linear current gain to 8A.

It is possible to design a quasi with very low crossover distortion, and I have made quite a good amp using 3055's but I would still recommend complementary pairs.

John
  Reply With Quote
Old 20th January 2014, 11:59 PM   #363
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Quote:
Originally Posted by john_ellis View Post
ON semi (Motorola) used to put the 2N3055 with the MJ15015 on a data sheet.
They still do. Just in case I went back to get their latest. They changed a few details, the 2N3055A is now 2N3055AG, but they kept the old Ft as min 0.8MHz and max 6MHz for both the 3055 and 15015, and then the 15016 still very different at min 2.2MHz and max 18MHz.

I wonder where the mistake is. Would they really make them so different (I mean the 15015 and 15016).?????

Quote:
Originally Posted by john_ellis View Post
But the recent 2N3055/MJ2955 data sheet gives both having an fT of 2.5MHz (min).
Quite true. And this is confusing and makes me think there is some mistake somewhere. Why 0.8 to 6 on one sheet and min 2.5 on an other, for the same 2N3055 part??? Which is correct?

Quote:
Originally Posted by john_ellis View Post
I suspect that the older MJ15015 devices were intended to match the second breakdown performance of the old, original RCA 2N3055 (later 2N3055H). This means that they probably did use a wider base with lower fT. This is probably why some people still think the 2N3055 is "not good" for hifi.
But the breakdown on the MJs is at 120V isn't it? The 3055H back then was meant for 100V.

Anyway, I think the recent 3055s should be just fine for hifi, as they would now have a higher Ft than others such as the MJ15003/4 for example, which are widely used for hifi.

Quote:
Originally Posted by john_ellis View Post
The epi base devices which have fT's around 3Mhz are closer. BUt rather than using MJ15015/6's for higher voltages (which may be slow) I'd suggest using MJ15003/4 or going straight to the MJ21193/4 which have 4MHz fT's and linear current gain to 8A.
This could be done for sure. It would no longer be a 3055 based amp, but there is nothing wrong with that. After all we're trying to squeeze all we can out of the old parts, and this can still apply to the newer ones.

This would allow going much higher on the rails. Might require a few recalculations though. The protections would surely need to be recalculated for the higher soa.

Quote:
Originally Posted by john_ellis View Post
It is possible to design a quasi with very low crossover distortion, and I have made quite a good amp using 3055's but I would still recommend complementary pairs.
Can you share your 3055 amp schematic? This might be interesting.

All in good fun!
  Reply With Quote
Old 21st January 2014, 12:05 AM   #364
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Now that I think of it, is there any way to measure the Ft???

Could that be built into a tester that's not too complex?

After all the manufacturers must have some way to test this. They would not just specify this in datasheets only from their calculations, they would do real tests.

Last edited by spookydd; 21st January 2014 at 12:17 AM.
  Reply With Quote
Old 21st January 2014, 06:32 PM   #365
diyAudio Member
 
Join Date: Sep 2006
You raise several issues.

(1) Ft is not too difficult to measure. Set up a class A bias circuit at the current you want to measure the fT at. Use a low impedance collector (non-inductive) load. Make or obtain an oscillator and if necessary build a high frequency, high current drive amplifier and feed the signal into the base using a resistor, or modulate the base current using a transistor bias circuit. Measure the change in base current and the change in collector current (i.e. ac. voltage across the current monitor resistor) to obtain the gain. If you make a spot measurement at say 1 MHz then the fT will be the ratio of the two directly. If you want to check that you are in the roll-off region measure the gain at several frequencies from 100kHz to 5 MHz.

(2) I agree that if we have to use higher voltage transistors we're not talking about 3055 amps any more. But the 3055 will work from 35 V rails and as mentioned most 3055's from ST or ON Semi will be happy up to 40V - always check your devices first though.

You could of course bridge two low voltage amps - say using a couple of pairs in parallel to get 100W into 8 ohms. Actually using a fully diff amp seems to improve second harmonic by common moding.

(3) Motorola tried for years to match the second breakdown of the RCA 2N3055(H). Their MJ15015 etc was probably based on developing a high second breakdown variant. ON Semi offer the 2N3055A which is supposed to match the 2N3055H and is probably built from the same line that makes the MJ15015. I suspect that the 2N3055 is from a different line but both devices will (I suspect) be built in the same lines as other transistors with 100V or more breakdown.

So the low frequency 2N3055A is probably from one line the 2N3055 from another.

(4) There are always differences between PNP and NPN transistors that make it impossible to obtain an exact match. But some of the modern devices seem to be very good.

(5) Nothing special about a low crossover quasi. Connect the stabilising capacitor inclusive of the output stage (Miller feedback makes things worse). I used a symmetrical configuration in terms of base resistances etc without a Baxandall diode. With the output stage included in the stabilisation compensation loop the equalising diode has little effect.

John
  Reply With Quote
Old 21st January 2014, 08:43 PM   #366
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Quote:
Originally Posted by john_ellis View Post
(1) Ft is not too difficult to measure. Set up a class A bias circuit at the current you want to measure the fT at. Use a low impedance collector (non-inductive) load. Make or obtain an oscillator and if necessary build a high frequency, high current drive amplifier and feed the signal into the base using a resistor, or modulate the base current using a transistor bias circuit. Measure the change in base current and the change in collector current (i.e. ac. voltage across the current monitor resistor) to obtain the gain. If you make a spot measurement at say 1 MHz then the fT will be the ratio of the two directly. If you want to check that you are in the roll-off region measure the gain at several frequencies from 100kHz to 5 MHz.
Thanks, so this is going to be something to consider in making a good tester.

Quote:
Originally Posted by john_ellis View Post
(2) I agree that if we have to use higher voltage transistors we're not talking about 3055 amps any more. But the 3055 will work from 35 V rails and as mentioned most 3055's from ST or ON Semi will be happy up to 40V - always check your devices first though.
That's the plan of course. And I am working on a bunch of leach amps as well and I wouldn't build any amp without testing and matching parts, including other important parts other than transistors.

Quote:
Originally Posted by john_ellis View Post
You could of course bridge two low voltage amps - say using a couple of pairs in parallel to get 100W into 8 ohms. Actually using a fully diff amp seems to improve second harmonic by common moding.
What we're working on here on this thread is a grounded bridge, so we'll have a good amount of power even at +-35V.

And when all parts are tested and properly selected, higher rails can be considered.

Quote:
Originally Posted by john_ellis View Post
(4) There are always differences between PNP and NPN transistors that make it impossible to obtain an exact match. But some of the modern devices seem to be very good.
With a large enough batch of parts, perhaps a good tester can allow making better selection and narrow the NPN and PNP's difference gap.

Quote:
Originally Posted by john_ellis View Post
(5) Nothing special about a low crossover quasi. Connect the stabilising capacitor inclusive of the output stage (Miller feedback makes things worse). I used a symmetrical configuration in terms of base resistances etc without a Baxandall diode. With the output stage included in the stabilisation compensation loop the equalising diode has little effect.
Intriguing. Can you share schematics?

I've been reading Bob Cordell's stuff recently and he's obtaining some good results, at least by calculations, on many topo improvements rather easy to implement and with cheap parts. There are probably some ideas to look at for our current project..
  Reply With Quote
Old 21st January 2014, 09:58 PM   #367
diyAudio Member
 
Join Date: Sep 2006
Oh all right then.

Features: "Cherry/Stochino" high input bias current to prevent cut-off during transients, cascode driver to eliminate Early effect distortion, high VAS current to prevent VAS cut-off during transient. Output stage quasi - excellent matching between halves, low distortion (but probably not as low as could be achieved with full complementary).

Effectively a Cherry-Blameless quasi combo!

John
Attached Images
File Type: jpg JNE50_Quasi.jpg (50.0 KB, 218 views)
  Reply With Quote
Old 21st January 2014, 10:37 PM   #368
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Interesting.

Quote:
Originally Posted by john_ellis View Post
Features: "Cherry/Stochino" high input bias current to prevent cut-off during
Doesn't high current there increase noise a lot?

Why is there an extra zener D5 near the current source?

Quote:
Originally Posted by john_ellis View Post
transients, cascode driver to eliminate Early effect distortion, high VAS current to prevent VAS cut-off during transient. Output stage quasi - excellent
Bob Cordell described such an improvement, with the benefits. What I don't understand is the purpose of C9, which must be what you described as "inclusive of the output stage", instead of the miller cap. How does that miller cap make things worse?

Quote:
Originally Posted by john_ellis View Post
matching between halves, low distortion (but probably not as low as could be achieved with full complementary).
That's ok, we know what we get when using quasi, but still, it's not that bad.

How low is the distortion?
What about TIM?

Have you actually built one for real or just ran sims?

Why not adding an extra pre-driver stage and 2 pairs of outputs?
Those have also good extra benefits, not to mention being able to drive heavier loads. Cordell also described this and it's all a plus to do so.

I am wondering about C1 and C7 though. Is the amp unstable without them?

What's the bandwidth obtained?
  Reply With Quote
Old 22nd January 2014, 06:46 PM   #369
diyAudio Member
 
Join Date: Sep 2006
Too many questions.
Yes I have built one and it sounds good. Distortion simulated is .007% at 20kHz. Bandwidth over 200kHz. If you use a Miller capacitor the VAS driver is slowed down at the very point you want it to speed up to minimise crossover distortion. This is the whole point! Yes it needs "sprog stopper" capacitors. Try to get hold of Cherry's "nested Differentiating Feedback" articles. This isn't but it uses Cherry's approach to stabilising inclusive miller feedback.

John
  Reply With Quote
Old 22nd January 2014, 08:40 PM   #370
diyAudio Member
 
Join Date: Apr 2012
Location: Florida & France
Quote:
Originally Posted by john_ellis View Post
Too many questions.
Sorry about that, I'm just curious and I had never heard of that "nested Differentiating Feedback" before.

Quote:
Originally Posted by john_ellis View Post
Yes I have built one and it sounds good. Distortion simulated is .007% at 20kHz. Bandwidth over 200kHz. If you use a Miller capacitor the VAS driver is slowed down at the very point you want it to speed up to minimise crossover distortion. This is the whole point! Yes it needs "sprog stopper" capacitors. Try to get hold of Cherry's "nested Differentiating Feedback" articles. This isn't but it uses Cherry's approach to stabilising inclusive miller feedback.
Ok cool, I'll dig something up.

Not bad result though. Probably too complex to attempt for the average diyer.
  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
the ultimate class A 2N3055 amplifier filippo22 Solid State 6 28th April 2014 12:02 PM
OP Amp based headphone amplifier. zebra100 Headphone Systems 32 1st October 2008 10:00 AM
LM3886 based amplifier russo Chip Amps 18 4th February 2008 12:27 PM
Dr. Amp Tripath based amplifier pepolman Class D 8 6th August 2005 07:28 AM
Op Amp Based Mosfet Power Amplifier Dusk Solid State 0 17th July 2005 07:13 PM


New To Site? Need Help?

All times are GMT. The time now is 07:10 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