The MONGREL (supersym II)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I saw that.
Self considers a dual transistor situation and compares voltage quality to the single transistor.
Many have used duals and report stability problems.
Some have used duals with no reported problems.
The consensus seems to be that a single will give rise to least problems.
Adopting the extra resistor is particularly good if there is only a small variation in the multiplier current.
 
you even forced me to draw it out.

There are two audio signal outputs from the Vbe multiplier.
The upper feeds the upper driver.
The lower feeds the lower driver.

NPN multiplier has the extra resistor in the top feed.
PNP multiplier has the extra resistor in the lower feed.

Whether it's a CCS or a bootstrapped resistor does not affect the two audio feeds to the driver stage.
Replace the lower CCS transistor with a complementary VAS. It looks exactly the same if you travel from +ve rail to -ve rail.

All the drive current flows from the top/VAS, down to the drivers. If we use a PNP multiplier with 10R on the bottom, most of the signal flows through C86 to get to the lower output. If we use an NPN multiplier, the signal must go through the 10R resistor, increasing voltage drop also across the multiplier resistor network, increasing conduction through the transistor. With a PNP multiplier, there is only 1 low-impedance path between the VAS and drivers.

- keantoken
 
All the drive current flows from the top/VAS, down to the drivers. If we use a PNP multiplier with 10R on the bottom, most of the signal flows through C86 to get to the lower output. If we use an NPN multiplier, the signal must go through the 10R resistor, increasing voltage drop also across the multiplier resistor network, increasing conduction through the transistor. With a PNP multiplier, there is only 1 low-impedance path between the VAS and drivers.

- keantoken
looks like we can't agree.
Any chance we could ask LTspice to adjudicate?
 
you even forced me to draw it out.

There are two audio signal outputs from the Vbe multiplier.
The upper feeds the upper driver.
The lower feeds the lower driver.

NPN multiplier has the extra resistor in the top feed.
PNP multiplier has the extra resistor in the lower feed.

Whether it's a CCS or a bootstrapped resistor does not affect the two audio feeds to the driver stage.
Replace the lower CCS transistor with a complementary VAS. It looks exactly the same if you travel from +ve rail to -ve rail.

Yes , the signals and voltages are the same on the 2 "sides" , but I see a big difference in the current through the 10R resistor between topologies. I simulated the 1/bootstrap - bx ,2/blameless - ax, and 3/symasym- cx at exactly 6ma- no signal current per amp. All 3 simulate differently using the same Vbe/OPS. ?? :confused: he he ..not really.
 

Attachments

  • Ic_compare.gif
    Ic_compare.gif
    34.4 KB · Views: 572
All the drive current flows from the top/VAS, down to the drivers. If we use a PNP multiplier with 10R on the bottom, most of the signal flows through C86 to get to the lower output. If we use an NPN multiplier, the signal must go through the 10R resistor, increasing voltage drop also across the multiplier resistor network, increasing conduction through the transistor. With a PNP multiplier, there is only 1 low-impedance path between the VAS and drivers.

- keantoken
That would be dependent on topology.. ??


os
 
First, all my comments are on the latest schematic, with a single-ended VAS. So it won't help for a balanced VAS.

The VAS is a high impedance output. This means that no matter it's output voltage, it will push whatever current is necessary in order to get the amp's output where it must be. How this current flows through the multiplier determines how well bias is held.

I've simulated this with the simplistic BAKSA circuit. The top traces are for the top circuit, bottom for bottom. As you can see, for the bottom arrangement most of the drive current flows through the cap. For the top one, current is split up through different components. Multiplier impedance is also less, as can be seen in that the voltage between driver bases is halved.

- keantoken
 

Attachments

  • DselfM.png
    DselfM.png
    63.2 KB · Views: 570
  • DselfMtraces.png
    DselfMtraces.png
    12.6 KB · Views: 561
First, all my comments are on the latest schematic, with a single-ended VAS. So it won't help for a balanced VAS.

The VAS is a high impedance output. This means that no matter it's output voltage, it will push whatever current is necessary in order to get the amp's output where it must be. How this current flows through the multiplier determines how well bias is held.

I've simulated this with the simplistic BAKSA circuit. The top traces are for the top circuit, bottom for bottom. As you can see, for the bottom arrangement most of the drive current flows through the cap. For the top one, current is split up through different components. Multiplier impedance is also less, as can be seen in that the voltage between driver bases is halved.

- keantoken

I see , I now know which to use on a dedicated blameless. Still, I must compromise , as I will end up with 4 balanced VAS's and 4 singles which all must work with the same OPS. This makes me wonder how the commercial "boys" and the top 3 amps on the forum get away with mje340's , "floating wipers" , and sometimes NO caps. Can you think of a good compromise ?

Oh.. a rush amp !! below..
OS
 

Attachments

  • bigrush.jpg
    bigrush.jpg
    81.2 KB · Views: 569
Oops, too bad OS, I think it's a false alarm (happens all the time with the Rush)! That is a very interesting design though... Can't have too much gain as evidenced by lots of degeneration, or the operating point is not stable. Conservative on NFB.

EDIT: Now I see Q11 and Q20, this is outside of the signal path so I didn't see it as a "Rush" at first.

To avoid the floating wiper issue, make R79 (post 314) the trimmer. This way a disconnected wiper will only decrease bias to 0. Or perhaps I am missing something?

- keantoken
 
Disabled Account
Joined 2006
Os, you can sim it if you have a model for such transistors, youll see that even the magic cap is not needed.

Theres no stability problems with darlingtons or a super beta transistor, it can happen if youre using a CFP in the multiplier but that only happens if your cap accross it is not big enough and other factors. The low voltage comes from transistors with low vce have higher beta, thats all, the maximum voltage accross a vbe multiplier will probably never exceed 10 volts so even a 20 volt tranny is good for the job.
 
Hi OS

Thanks for developing the PB60LP and answering my questions the other day.

The PB60LP now looks complete on your website, but i did spot a minor ommision on the PCB layout - the through-holes are missing for the output quick-connect. Obviously this is very minor but just bringing it to your attention.

Craig
 
Hi OS

Thanks for developing the PB60LP and answering my questions the other day.

The PB60LP now looks complete on your website, but i did spot a minor ommision on the PCB layout - the through-holes are missing for the output quick-connect. Obviously this is very minor but just bringing it to your attention.

Craig
Have to post again... freakin' moderation happy fools.
Here is the correction, craig , small as it may seem.(pix1)
You can use these with the 2 outer holes.
19705notabiso.jpg


or drill the inner hole out bigger and use my favorite , this ..
prx528040b.jpg

The correction and my new board PBM200 (2X IRF240/9240) is below.
Homemodder , you have mosfet experience , a few tips.. I HAVE built the CX voltage stage with IRF240/9240 , but any tips/suggestions would be appreciated... maybe even the super beta as Vbe ??
I promised not to post in carlo's threads , I do not ignore people , I just want better than "junk" with no adaptability.
OS


OS
 

Attachments

  • correction.gif
    correction.gif
    36.9 KB · Views: 513
  • pbm200.gif
    pbm200.gif
    35.7 KB · Views: 512
  • pbm250_schema.gif
    pbm250_schema.gif
    109.1 KB · Views: 535
Disabled Account
Joined 2006
Hopefully we can get back now to what was being discussed.

Os I was thinking of asking keantoken to sim a vbe multiplier using a darlington and make the same comparison he did in the aksa thread. Im trying to find transistor models for the darlington at the moment. I dont really know what to say about the outputstage, In all the years I still havent built anything with quasy outputstages.
 
According to B. cordell , the quasi is inferior anyway with the better availability of P channel devices. He said it is better to accept the small differences in P/N channel devices than to use the totally unbalanced quasi topology anyways. Below is the AX "blameless" ( sim and pix) with the 2X IRFP240/9240 PB200. I almost have it "close" to the BJT blameless. (.0008% 1kHZ)... could be a model thing. :( Vbe is stable to a few mA WRT freq. and power level .. still it is very sensitive.

OS
 

Attachments

  • AX_PBM200.gif
    AX_PBM200.gif
    23.7 KB · Views: 326
  • mongrel_AX1.0_mosfet.zip
    8.5 KB · Views: 114
According to B. cordell , the quasi is inferior anyway with the better availability of P channel devices. He said it is better to accept the small differences in P/N channel devices than to use the totally unbalanced quasi topology anyways.

I remember reading somewhere that Bryston has an output triple/CFP which does well in hiding differences between N/P types

quote "the complementary characteristics of the upper and lower halves of this output configuration are almost ideally matched"

http://bryston.com/chrismemo/index.html

and see D. Self's book, chapter 6, p156, fig 6.17c
 
I was informed of my "wrongspeak" in previous posts and make a public apology for any undue criticism of the moderation. I would even extend a hand to "uncle charie" if he would do as I do now.

You got it right and fully optimised this time, it will bring out the best performance out of those fets. Some say the quasy sound very good, I guess I should still try this one day but I haven,t found anything wrong with the complimentary setup.

Thanks to you , bro. I read your comments in another thread as well as cordells comments. Would .15 / .33R source Re's be even better to compensate for the mismatch ??
OS
 
I remember reading somewhere that Bryston has an output triple/CFP which does well in hiding differences between N/P types

quote "the complementary characteristics of the upper and lower halves of this output configuration are almost ideally matched"

Bryston Limited - Music For A Generation

and see D. Self's book, chapter 6, p156, fig 6.17c

Bryston marketing hype !!:D
OS
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.