Comments on unusual bridge amp?

Status
Not open for further replies.
Jump to Latest
#22
Commentable Experienced Thoughts

Hi Doug
See this so u can get familiar with feedback
After this we will have something for u.

hope that it helps
 

Attachments

  • diff.jpg
    diff.jpg
    12.3 KB · Views: 547
#23
Anyone wanting to build a bridge amp should look at these two old workhorses:

http://home.kimo.com.tw/skychutw/

Then click on 'circuits'

Then right click and save 'Hadley 622C'

Then right click and save 'Liquid-Cooled Power Amplifier'

Why are these of interest to us?

One has a quasi-comp output stage, the other is full-comp. The point is you could make either what ever way you want.

They both show the 'X' feedback scheme and could be adapted to 'Suzy'

The 'Hadley 622C' only has one voltage gain stage, what is basically the Vas is the only voltage gain stage for the whole amplifier.

The 'Hadley 622C' shows electronic current limiting, the 'Liquid-Cooled Power Amplifier' shows a simple fuse set up.

With careful selection of modern BJTs, 400W at 4 ohms is possible from only one set of inexpensive transistors.
 
#24
Hi djk,

Thanks for the webpages. They do look interesting. I will try to see if I can understand the X- bridge feedback designs.

The Hadley amp look like it uses lots of different power supplies so I find it rather unelegant. And the first phasesplitter stage would (I think) cause a BIG thump at startup while the coupling caps charge. And because of these caps that you cant linearize this stage with feedback. And the 2 transistors in the middle of the bridge (Q104? and Q34?). I cant seem to figure out what these are supposed to do. My neck is getting tired from looking at the schematic ;-)

I think you are wrong about the water cooled amp having only 1 voltage gain stage. The power transistors are common emitter in the diagram so they would give voltage gain too. But I'm not convinced that the curcuit would even work as shown. The output stage biasing looks very odd. The resistors in the collectors of the LTP would also lower the PSRR.

Doug
 
#26
"The Hadley amp look like it uses lots of different power supplies so I find it rather unelegant."

So don't build it that way. We are interested in the concepts these amps can show, not the minutia (all the semiconductors are probably obsolete anyway).

" And the first phasesplitter stage would (I think) cause a BIG thump at startup while the coupling caps charge."

So what? don't you have a preamp or CD player with a balanced output to drive it with?

" And because of these caps that you cant linearize this stage with feedback. "

We weren't going to build this stage anyway, but really, its a class A unity gain stage, how much more feedback do you think it needs?

"And the 2 transistors in the middle of the bridge (Q104? and Q34?). I cant seem to figure out what these are supposed to do. My neck is getting tired from looking at the schematic ;-)"

I rotate the schematic so I don't have to crane my neck. Those are part of the current limiter (see the 0.063 ohm resistors?)

"I think you are wrong about the water cooled amp having only 1 voltage gain stage"

I said:

"The 'Hadley 622C' only has one voltage gain stage,"

and:

"The 'Hadley 622C' shows electronic current limiting,"

"The power transistors are common emitter in the diagram so they would give voltage gain too"

They are wired for unity gain, 100% feedback.

"The output stage biasing looks very odd. "

What's odd about a straight voltage drop across a resistor? Its water cooled and biased about into class A.

"The resistors in the collectors of the LTP would also lower the PSRR."

Without the voltage drop across those resistors you would have nothing to drive the Vas with. What is confusing you is that with those values the diff pair have NO voltage gain, they are just producing the drive signals for the Vas stages.

Maybe this is more to your liking:

http://users.ece.gatech.edu/~mleach/lowtim/instage.html

The LC Audio End Millenium also uses very low gain in the diff pairs.

"I will try to see if I can understand the X- bridge feedback designs."

The best thing to do here is to download the text of 5,376,899 and closely read BACKGROUND ART fig. 7, 8, 9 while looking at your downloaded PUBLCATION IMAGES pg, 6, 7 fig. 7, 8, 9

Fig 7 is a Sumo Model 9 feedback scheme, fig. 8 is Sandman (the Yamaha B2X uses this design), and fig. 9 is the patent. Note that the basic feedback loops of the 'Hadley 622C' and the 'Liquid-Cooled Power Amplifier' have the basic 'X' feedback scheme (the 'X' being easiest seen in the 'Liquid-Cooled Power Amplifier' )

What they lack are "positive input terminals are coupled to the corresponding outputs in a way that allows the two stages to negatively feed each other the differential signal so that it is reinforced at the outputs and to positively feed each other their own distortion and noise contributions so that they are cancelled at the outputs. " (from the patent text). This is shown as resistor R166 in pg. 7, fig. 9
 
#33
It seems to me that a string of two well chosen diodes in the VAS collector between the output drivers would reduce the crossover region and bring the outputs closer to the AB threshold, without actually going there and requiring true bias servo circuits.
 
#36
Not yet 🙁 have been distracted with work and house renovations for the new baby 🙂))))

http://www.deleveld.dds.nl/bridge8d.gif

I changed the VAS current sources to have negative resistance which is maintained in the crossover region. From simulations this seems to slightly reduce crossover distortion. It means some degree of positive feedback exists but as long as it isnt too much then there (hopefully) wont be problems with stability. For symmetry I would have liked to have the VAS top also as a differential pair but this gives too much gain for the positive feedback portion and seems to have some stability problems close to the crossover region (where the negative feedback doesnt work).

I also changed the input configuration to something I think is rather clever. The new input stage has high input impedence, which was a problem with the old topology. Also the DC level at the output is spread across the feedback resistors and the input transistors. For my example it is spead evenly but it is possible to have only a few volts across the C/E of the input pair regardless of the supply voltage. This makes it feasible to use low-voltage super high beta transistors for the input stage, even for high supply rails. They can even be run at high currents for very high gm. I would like to use MAT here but I cant seem to find a supplier I want on the web. The 2N3904 seem fast enough with a good beta, very cheap and available everywhere. And they are in LTSpice.

The final version wont use 2n3055 in the output stage, these are is the schematic only because its the only high power transistor in LTspice. I'm looking at 2sb1560/2sd2390 here. These darlingtons are fast and reasonably cheap. And I like the idea of a low parts count. I found them on the web at BMM electronics and may do an order soon.

Try a simulation of the topology. I think youll find that it (should be) stable and has a very good PSRR. Dont let r13 and r17 confuse you, they only supply current for the zener and the cap. I could have used a resistor from posirtive rail instead, but this seems more elegant for my initial parts layout.

Doug
 
Status
Not open for further replies.
Top Bottom