Leach Amp different emitor resistors

[nigelwright7557]

Originally Posted by Borc

I intend to change emitor resistors on output devices from default 0,33 Ohm to 0,22 Ohm (Mills group buy). I removed protection circuit so this is not an issue.
I would like to have some help on the new optimal bios for output devices. Is there any mathematic formula to determine it.
For the same voltage drop on emitor resistors this will be 67,5mA
(from default 45mA),

Regards

I always set up bias with a scope, signal generator and a 4 ohm speaker connnected.
I apply a 1 volt sine wave and slowly increase the bias until crossover distortion goes. Any less bias gives crossover distortion and any more bias just wastes heat in the heatsink.

I alwasy use 0r22 resistors. If you go too low one pair of transistors will be taking all the load.

[djk]

So a Carver M1.5 isn't very stable with it's 0R05 emitter resistors?

[satoru]

Originally Posted by djk

So a Carver M1.5 isn't very stable with it's 0R05 emitter resistors?

The size of RE is dependent on the Vcc, type of transistor (Rbb' or the size of internal emitter ballast registers, thermal resistance from chip to case), size of heatsink, etc. By the way, out of curiosity, I downloaded the PDF file of the manual which contains the schematics and I see 3W 0R1 resistors there. Is your RE value coming from a real installation? Or did I read the schematics wrongfully??

A rough (and crude) way to calculate the minimum size of RE to prevent thermal runaway is:

RE ((+ internalRE)) > Rdjc * Vcc * 0.002 (V/°C) (( - 1/gm))

This equation is for an installation that the bias compensation transistor (or diode) is mounted directly on the output transistor with minimum thermal resistance in between. The double bracket terms are optional for calculation. Rdjc is the junction to case thermal resistance, Vcc is half of the total supply voltage (+/-Vcc) for push-pull output stage, and 0.002 is rough approximation for the thermal coefficient of silicon transistor. internalRE is Rbb'/hFE or internal ballast RE. Rdjc is 0.7 °C/W for MJ15024/MJ15022. So, for a Vcc=80V (+/- 80V total power supply rail) amplifier with MJ15022/MJ15024 pairs, 0R05 RE is OK, but 0R10 is safer. Smaller RE results in less distortion from the output pair while raises more concern for the thermal runaway in extreme conditions.

Regards,
Satoru

[Tony]

Satoru-san,

can you post links to manuals please?

thanks,
tony

[AndrewT]

Originally Posted by satoru

The size of RE is dependent on the Vcc,.........
.......(Rbb' or the size of internal emitter ballast registers........
........ the minimum size of RE to prevent thermal runaway is:

RE ((+ internalRE)) > Rdjc * Vcc * 0.002 (V/°C) (( - 1/gm))......

what's the significance of the minus (-) in the -1/gm?
What is a valid range of gm for output BJTs?

Does the output base stopper, if fitted, add to Rbb? or some other correction?

This formula for thermal stability seems very similar to Bob Cordell's. Are they from the same source?
Cordell says that the internal Re and the gm vary with bias current. Does your version do similar?

[satoru]

Originally Posted by Tony

Satoru-san,

can you post links to manuals please?

thanks,
tony

Dear tony-san,

This formula is taken from Toru Kuroda's book (in Japanese), Basic Design of Transistor Amplifiers, 1989. It is a compilation of his series appeared on a Japanese DIY magazine. Sorry but it's long out of print.

Originally Posted by AndrewT

what's the significance of the minus (-) in the -1/gm?
What is a valid range of gm for output BJTs?

Does the output base stopper, if fitted, add to Rbb? or some other correction?

This formula for thermal stability seems very similar to Bob Cordell's. Are they from the same source?
Cordell says that the internal Re and the gm vary with bias current. Does your version do similar?

Dear Andrew,

Unfortunately, base stopper won't reduce the chance of thermal runaway at the values commonly used, as I understand. Please correct me if I'm wrong (as often it is the case).

A value of gm for a modern power transistor in high current range is around 30S (our favorite 2SC5200 at 25°C , roughly estimated from a plot in data sheet). It gets lower as the temperature goes higher. Not a big contribution to the RE value calculation but comes into scene when we are talking about RE value around 0.05 ohm. In his book, Kuroda removed those terms which have dependencies on bias current. I included them since the discussion was on lower end of RE values. Oh, the 1/gm term should be on the left conceptually, but I moved it to right (with minus, which mathematically the same but wrong in terms of physics).

I'll check Bob Cordell's paper. Kuroda didn't cite any papers but laid out details on his model and calculations. It is very likely that he used the same source as Bob's formula or even Bob's writing.

Omitting RE has one big benefit: lower distortion from output stage. The risk is obviously the thermal runaway. Kaneta (a famous DIY designer almost forming a cult group around him in Japan) published many designs with no RE. He specified older generation output transistors for his circuits and those are more vulnerable to thermal runaway. As a result, you can find a lot of tragic stories online. Kaneta amp fails and drags down the speaker unit together. So many and too many of them.

Kuroda wrote in the preface of his earlier book that the motivation of writing an introductory design book was the anger against "practice of some DIY designers releasing circuits based on wrong concepts and sometime even dangerous designs." Kuroda didn't say who did so, but after reading many articles in DIY magazines in Japan, I came to realize that it could well be that Kuroda was pointing toward Kaneta's designs.

All corrections and comments are welcome!

Regards,
Satoru

[pooge]

Originally Posted by satoru

A rough (and crude) way to calculate the minimum size of RE to prevent thermal runaway is:

RE ((+ internalRE)) > Rdjc * Vcc * 0.002 (V/°C) (( - 1/gm))


Regards,
Satoru

Your equation is mathmatically unclear or inaccurate, especially since you used parenthesis around terms in a non-mathematical way. If, as you stated in a later post, the 1/gm term was moved from the left, I assume it should be subtracted from the entire right side term as below?

RE + internalRE > [Rdjc * Vcc * 0.002 (V/°C)] - 1/gm

Also, how is the internal RE determined, as it is variable?

[satoru]

Originally Posted by pooge

Your equation is mathmatically unclear or inaccurate, especially since you used parenthesis around terms in a non-mathematical way. If, as you stated in a later post, the 1/gm term was moved from the left, I assume it should be subtracted from the entire right side term as below?

RE + internalRE > [Rdjc * Vcc * 0.002 (V/°C)] - 1/gm

Also, how is the internal RE determined, as it is variable?

Ah, now I see my mistake. The equation should be

RE + 1/gm + internalRE > Rdjc * Vcc * 0.002

I tried to move the 1/gm to right to make it clear how RE is determined and made confusions...sorry. You can estimate the value of internalRE at a certain bias point but since it is variable, usually it is omitted from the calculation. In the case of ballast RE, there is minimum value, which in turn, can be used in the calculation safely.

Regards,
Satoru

[satoru]

OK, call me a fool, shame on me. I tried to mix two equations from two different sections of the book and messed up big. Following equation is straight from the book and should be regarded (not my messed up ones appear in this thread).

RE + 1/gm > Rdjc * Vcc * 0.002

If you ignore the 1/gm term, then you get a RE value for very "safe" operation.

RE > Rdjc * Vcc * 0.002

Thanks all for pointing out my mistakes.

As I understand, there is no away my posts containing serious mistakes to be removed but hope it'll serve a lesson to myself and some potential beginners...

Regards,
Satoru