I've created a bunch of problems for myself.
Trying to increase drive currents in an insane
effort to overwhelm 1500pF extra Miller. That
Bob Cordell warns will happen when VDS<VGS.
Anyways: Due to my changes, input impedance
was getting silly low between the emitters, and
something had to be done. And my fix for that is
really all this particular posting is about.
The rest of the circuit is probably way off, and
likely to be completely different before too long.
The soft current limiter rings!!! and I don't know
how best to stop it... Then again, it only rings
into 2ohm loads and less. I need the hard limit
to take over before this happens...
I'm thinking input transistors of the same temp
in an MPQ DIP might give the lowest offset?
Trying to increase drive currents in an insane
effort to overwhelm 1500pF extra Miller. That
Bob Cordell warns will happen when VDS<VGS.
Anyways: Due to my changes, input impedance
was getting silly low between the emitters, and
something had to be done. And my fix for that is
really all this particular posting is about.
The rest of the circuit is probably way off, and
likely to be completely different before too long.
The soft current limiter rings!!! and I don't know
how best to stop it... Then again, it only rings
into 2ohm loads and less. I need the hard limit
to take over before this happens...
I'm thinking input transistors of the same temp
in an MPQ DIP might give the lowest offset?
Attachments
Last edited:
The MPQ series would give the ultimate temperature coupling, but I don't think they would be specially matched. Maybe a little more reliably similar because of being made at the same place, presumably on the same machine. A look at OnSemi came up with this:
http://www.onsemi.com/pub_link/Collateral/MMPQ6700-D.PDF
EDIT: I just noticed the big text saying "obsolete". I wish they would suggest a replacement!
- keantoken
http://www.onsemi.com/pub_link/Collateral/MMPQ6700-D.PDF
EDIT: I just noticed the big text saying "obsolete". I wish they would suggest a replacement!
- keantoken
There is also this:
http://www.onsemi.com/pub_link/Collateral/BC846BPDW1T1-D.PDF
With half the Ft of the MPQ6700...
- keantoken
http://www.onsemi.com/pub_link/Collateral/BC846BPDW1T1-D.PDF
With half the Ft of the MPQ6700...
- keantoken
Here we trade respectable Ft for unpredictable current gain.
http://www.onsemi.com/pub_link/Collateral/EMZ1DXV6-D.PDF
- keantoken
http://www.onsemi.com/pub_link/Collateral/EMZ1DXV6-D.PDF
- keantoken
This is looking good:
http://www.onsemi.com/pub_link/Collateral/MBT3946DW1T1-D.PDF
My recommendation.
- keantoken
http://www.onsemi.com/pub_link/Collateral/MBT3946DW1T1-D.PDF
My recommendation.
- keantoken
Last edited:
I don't need NPN temps to track PNPs. That
package style just gives me four pairs that
still may not be at the same temp.
I just need all four NPNs to match each other.
And then all four PNP to match each other...
And then all four offset pairings will be equal.
VBE is the only parameter that need match.
--------------------
Its the same circuit as with four JFETs
except the matching is assured by VBE
with cheap bipolar transistors. Able to
deal with higher voltages and currents.
package style just gives me four pairs that
still may not be at the same temp.
I just need all four NPNs to match each other.
And then all four PNP to match each other...
And then all four offset pairings will be equal.
VBE is the only parameter that need match.
--------------------
Its the same circuit as with four JFETs
except the matching is assured by VBE
with cheap bipolar transistors. Able to
deal with higher voltages and currents.
Last edited:
I understand complimentary matching is not necessary. I still don't think transistors of the same polarity in the same package will be necessarily well matched unless such is indicated on the datasheet. Perhaps a little better than discrete devices though since they come off the same machine.
- keantoken
- keantoken
Allison A/B update
Keantoken,
It's been a while. I've finally gotten around to building an interface for my sound card to run the Allison A/B at full power. I built Pete Millett's interface, a very nice device! I found that after running the V6a for the past year that it wasn't doing so well. I swapped out the fast toshiba drivers for some 15032/33 and things look good again. My prior measurements were always at low power. So, now I have measurements at about (in the following order) 2vRMS into 8 ohms, 4vRMS and 17vRMS. The 16vRMS is just before clipping sets in. I've also switched software to Audio Tester. The thd numbers are snapshots with plus minus variations of .0005 or so as you watch it cycle. Same for the graphs. Oops, i've exceeded file size limits. Will redo and attach to next post.
Ken L
Keantoken,
It's been a while. I've finally gotten around to building an interface for my sound card to run the Allison A/B at full power. I built Pete Millett's interface, a very nice device! I found that after running the V6a for the past year that it wasn't doing so well. I swapped out the fast toshiba drivers for some 15032/33 and things look good again. My prior measurements were always at low power. So, now I have measurements at about (in the following order) 2vRMS into 8 ohms, 4vRMS and 17vRMS. The 16vRMS is just before clipping sets in. I've also switched software to Audio Tester. The thd numbers are snapshots with plus minus variations of .0005 or so as you watch it cycle. Same for the graphs. Oops, i've exceeded file size limits. Will redo and attach to next post.
Ken L
Hello Klewis, long time no see! I have learned many things since we last talked.
Too bad about the stability issues. I don't remember which schematic this is from, do you have it on hand?
I'm smarter and more knowledgeable now, if you're looking for a Round Two. I will look at the old schematics again if I can find them, there may be new tweaks to make them more stable.
- keantoken
Too bad about the stability issues. I don't remember which schematic this is from, do you have it on hand?
I'm smarter and more knowledgeable now, if you're looking for a Round Two. I will look at the old schematics again if I can find them, there may be new tweaks to make them more stable.
- keantoken
keantoken,
yeah, it's been a while. The stablity issue is from the 2sc4793/1837's when swapped out I get the output shown. The schematic was version 6a, the posts start around #380. I looked over the posts a couple of days ago and had forgotten how much work had been put into this arrangement. I seem to remember that it runs best outside of the feedback loop for the 49811, though I will go back this weekend and try the current arrangement with the feedback wrappped around the whole thing.
It is running well with the current arrangement, but, tweaks are welcome. As you can see there is a small spike at 2k and 3k when run at full output.
My next plans are to run the 49811 on a Salas v1.2 shunt reg at a higher voltage and see what happens.
Ken
Good idea in many ways! Running the Allison and LME on different rails should aid stability, since their parasitic resonances won't interact.
I can't find the exact schematics, I wonder if they even exist. LOL! There were a few basic types though. Does the one you're discussing use CFP or Darlington outputs? Is it class A or class AB?
- keantoken
I can't find the exact schematics, I wonder if they even exist. LOL! There were a few basic types though. Does the one you're discussing use CFP or Darlington outputs? Is it class A or class AB?
- keantoken
Hey, sorry for getting back late. For some reason the forum doesn't notify me of some posts. I should complain about that.
Try putting 10R resistors in series with the bases of Q3 and Q4. If this works, work down until you find the lowest values that prevent oscillation, and move one value up.
If that doesn't work, put 100R resistors in series with the bases of Q1 and Q2, with the previous resistors in place (if this method works, you can test removing the resistors from Q3 and Q4). Repeat the procedure used for Q3 and Q4.
How has it sounded?
- keantoken
Try putting 10R resistors in series with the bases of Q3 and Q4. If this works, work down until you find the lowest values that prevent oscillation, and move one value up.
If that doesn't work, put 100R resistors in series with the bases of Q1 and Q2, with the previous resistors in place (if this method works, you can test removing the resistors from Q3 and Q4). Repeat the procedure used for Q3 and Q4.
How has it sounded?
- keantoken
keantoken,
I removed the 4793/1837s and put in 15032/33s. It runs great with these devices. As you might recal, the 4793/1837s were always trouble. If there is a performance reason werein the amp would benefit from there use, I can try again with them and the mods that you suggest. Your thoughts?
As for sound, it's sweat, I like it more than the EF configuration. Seems smoother and cleaner.
Ken
I removed the 4793/1837s and put in 15032/33s. It runs great with these devices. As you might recal, the 4793/1837s were always trouble. If there is a performance reason werein the amp would benefit from there use, I can try again with them and the mods that you suggest. Your thoughts?
As for sound, it's sweat, I like it more than the EF configuration. Seems smoother and cleaner.
Ken
Now that I think about it, there won't be much to gain from using the 4793/1837, since this is the AB design. So I wouldn't worry about it unless you were the "transistor roller" type.
Regarding the 15032/33 pair, it may be better to use the 15034/35 for class AB because of the better Hfe matching. It doesn't matter for class A so much:
http://www.diyaudio.com/forums/solid-state/169590-mongrel-supersym-ii-13.html#post2314634
I'm glad to hear about the sound, this is good news.
- keantoken
Regarding the 15032/33 pair, it may be better to use the 15034/35 for class AB because of the better Hfe matching. It doesn't matter for class A so much:
http://www.diyaudio.com/forums/solid-state/169590-mongrel-supersym-ii-13.html#post2314634
I'm glad to hear about the sound, this is good news.
- keantoken
kean,
After a bit of work I've made a seperate supply for the 49811 using Salas Shunt v1.2R at 60v out with the Allison running on 32v seperate supply.
A nice bit of improvement. Also, the sound has more bass and better attack! Quite nice.
Graphs of the sound card loop back, previous configuration, the 49811 alone and the 49811 driving the Allison 9.6v out. Actually saw thd of the 49811 alone at .0012, basically the noise floor of the rig.
Attachments
more questions
Keantoken
Since you're waiting for parts on the Goldmund, I thought I would bug you with a few questions.
So, I thought that I would try to compare a "normal" typology against the Allison A/B with the simulator. My intent is to see how a normal typology compares at different input voltages, and different frequencies. All of the images are run at 2v in and 1kHz.
The first image is with the simulator setup you suggested a long time ago.
The second image is with the simulator setup that Bob Cordell suggested for testing thermal trak bais with a normal typology.
The third image is the same simulator setup as no. 2 but with the Allison A/B.
My question is: is the third image represenative of the Allison being driven by the a quality VAS stage such as the LME49811?
Ken L
Keantoken
Since you're waiting for parts on the Goldmund, I thought I would bug you with a few questions.
So, I thought that I would try to compare a "normal" typology against the Allison A/B with the simulator. My intent is to see how a normal typology compares at different input voltages, and different frequencies. All of the images are run at 2v in and 1kHz.
The first image is with the simulator setup you suggested a long time ago.
The second image is with the simulator setup that Bob Cordell suggested for testing thermal trak bais with a normal typology.
The third image is the same simulator setup as no. 2 but with the Allison A/B.
My question is: is the third image represenative of the Allison being driven by the a quality VAS stage such as the LME49811?
Ken L
Attachments
In retrospect, that is in fact a perfect VAS. Since it has no impedance (as opposed to a normal VAS which generally has high impedance) stability will be much different and VAS current won't factor into the distortion. So it should not be taken as an indication of the distortion performance of that circuit.
The same is true for the first schematic. Performance within a feedback loop will depend on performance without. It is more better to analyze an output stage without a feedback loop.
- keantoken
The same is true for the first schematic. Performance within a feedback loop will depend on performance without. It is more better to analyze an output stage without a feedback loop.
- keantoken