In his amplifier book near the end of the section on thermal compensation , D.Self describes using a Sziklai pair for the Vbe multiplier. Apart from the possability (probability) that it would oscilate, it has the advantages of very low bias drift and elimination of the trimpot. Self doesn't seem very excited about his own idea.
On Hagerman's website (www.hagtech.com), Hagerman has an article about a "Vbe doubler" with somewhat similar properties although it is not a Sziklai and presumably doesn't have the same oscillation problems.
Has anyone tried to use either of these on a real amp or are they just paper curiosities? Did it work or were there difficulties?
Do either or both of them still require physical contact with output device or heatsink (or driver if CFB)? If yes, do mount both transistors or just one (which)?
If either of these are effective I'm surprised not to have seen anyone attempt to use them.
On Hagerman's website (www.hagtech.com), Hagerman has an article about a "Vbe doubler" with somewhat similar properties although it is not a Sziklai and presumably doesn't have the same oscillation problems.
Has anyone tried to use either of these on a real amp or are they just paper curiosities? Did it work or were there difficulties?
Do either or both of them still require physical contact with output device or heatsink (or driver if CFB)? If yes, do mount both transistors or just one (which)?
If either of these are effective I'm surprised not to have seen anyone attempt to use them.
The circuit that is on the Hagerman web site has been used by Harmon Kardon and a few others. It still requires that one of the transistors be attached to the heatsink.
Regards,
Jam
Regards,
Jam
Well, looks like the only way to satisfy my curiosity is to try on or another in my next amp project. It looks like I could layout the standard Vbe multiplier and add a couple of componrnts to the PCB so that I could first use a conventional circuit and when I'm satisfied that's stable and working, change component values and add the second transistor. That way I won't be puzzling whether it is the new circuit or something else that causes any problems.
And I want a cherry pie ... now
Mate, use the search button / google is your friend 😉
Sam9
---In his amplifier book near the end of the section on thermal compensation , D.Self describes using a Sziklai pair for the Vbe multiplier. Apart from the possability (probability) that it would oscilate, it has the advantages of very low bias drift and elimination of the trimpot. Self doesn't seem very excited about his own idea.---
An important advantage of the Sziklai pair for bias control is the lower dynamic impedance compared to a single transistor.
---In his amplifier book near the end of the section on thermal compensation , D.Self describes using a Sziklai pair for the Vbe multiplier. Apart from the possability (probability) that it would oscilate, it has the advantages of very low bias drift and elimination of the trimpot. Self doesn't seem very excited about his own idea.---
An important advantage of the Sziklai pair for bias control is the lower dynamic impedance compared to a single transistor.
Member
Joined 2009
Paid Member
I believe MJL21193 has used this type of multiplier on his Class A 'Nameless' amplifier.
http://www.diyaudio.com/forums/showthread.php?t=145476&highlight=feedback+Class
Scroll down to the schematic in Post 52. He has built it but hasn't had time to debug it yet.
Member
Joined 2009
Paid Member
It's a good question - but I think it's not hard to make it stable if there's a problem. This is the bias generator so we don't care about crippling it with some capacitance. In simulations, it's sufficient if you include a little emitter degeneration when using it with gain.
I've seen Sziklai circuits used in the multiplier in guitar amps for ages. Roland's Jazz Chorus 120 is one example that's been out there already from the early 1970's. If I remember correctly, many Gallien-Krueger amps also used a Sziklai.
It's not nearly as stable circuit as the single-transistor VBE multiplier. In fact, it's terrible. However, proper device choice can help making it more stable as well as Miller cap for high frequency negative feedback.
It's not nearly as stable circuit as the single-transistor VBE multiplier. In fact, it's terrible. However, proper device choice can help making it more stable as well as Miller cap for high frequency negative feedback.
If you dont want to run into stability problems just use a very high beta single transistor or even better a darlington, performance is very close to a CFP.
- Status
- Not open for further replies.