Slewmaster - CFA vs. VFA "Rumble"

[jkuetemann]

There's a cool little 'feature' with the CFA-XH that I wonder if anyone else has noted. You get a free clipping indicator! The Hawksford Cascode LEDs are starved of current just as clipping sets in and go out.

 

[still4given]

Hi OS,

It looks like you drew that schematic in spice. Any chance you would share the file. I can't seem to get mine working.

 

[5th element]

There's a cool little 'feature' with the CFA-XH that I wonder if anyone else has noted. You get a free clipping indicator! The Hawksford Cascode LEDs are starved of current just as clipping sets in and go out.

Hmm that's interesting. I could perhaps use that somehow

 

[ostripper]

Are R17 & R18 then necessary? They are part of the 'VAS abuse' compensation scheme, were they not? I have been testing the IPS in isolation without them (or the shunt Cs) and it seems to work without them.

150K is just a little "abuse" .
They change H2-H3 ratio and moderate VAS impedance changes. I noticed
the OEM's included them and had to see why ....

THD wise (on a EF3) , the hawksford simulates slightly lower with a tiny
load (150k-220k) ... also, this lowers CLG by a few DB.

Terry .. (below is the CFA-Xh .asc) ...

OS

 

[ostripper]

There's a cool little 'feature' with the CFA-XH that I wonder if anyone else has noted. You get a free clipping indicator! The Hawksford Cascode LEDs are starved of current just as clipping sets in and go out.

Devices saturate and 'starve" the led's - cool !
The simulation also shows this ... but ,Clipping is still smooth.

OS

 

[still4given]

Hi OS,

That did the trick. No more oscillation and playing beautifully.

As for the UV's working as a clipping indicator? You have better have some pretty impressive speakers and a big room. I don't think I could get this to clip with my B1. Bet you need over 2V input to get there.

Hey OS, maybe you should edit post #1 to direct folks to the latest schematic so they won't have to hunt for it.

Oh yeah, how important are the 3K9 resistors? I have the 5K6 per the earlier schematic. Do I need to change those.

Here's a pic of it playing happiliy along.

Blessings, Terry

 

[thimios]

Hi OS,

That did the trick. No more oscillation and playing beautifully.

As for the UV's working as a clipping indicator? You have better have some pretty impressive speakers and a big room. I don't think I could get this to clip with my B1. Bet you need over 2V input to get there.

Hey OS, maybe you should edit post #1 to direct folks to the latest schematic so they won't have to hunt for it.

Oh yeah, how important are the 3K9 resistors? I have the 5K6 per the earlier schematic. Do I need to change those.

Here's a pic of it playing happiliy along.

Blessings, Terry

Congratulations Terry!
Now it's time for comparison.
Can you compare this with PMI CFA?(same volume settings)
Thimios.

 

[still4given]

Hi Thimios,

Yes I can do that. Tomorrow morning sometime. I have a question for you. In your screen shots of the square waves at 100mHz and above, what kind of input cap are you playing through? The only way I have been able to get one to look like that is with the input cap bypassed.

Thanks, Terry

 

[thimios]

Right. First of all, this helps to confirm what both I and triplej suspected. At least partially.

I do not know if you have read Self, thimios, but there are a few reasons why performance like this can occur.

Basically what we're looking at is low distortion with the sound card in loop-back mode and then distortion that looks almost identical with the amplifier working without a load. Things then turn sour when the load is connected. This isn't necessarily a surprise, as measuring power amplifier can be quite a challenge.

As the distortion is low without a load connected we can assume that there is no non linearity being cause by resistors or capacitors within the voltage divider to the PC and the feedback within the amplifier itself.

When you connect a load to an amplifier this causes several things to happen all of which occur because current is now having to flow within it.

The first issue is that as current is drawn from the supply rails it causes them to become modulated to the tune of what is imposed on the output. When a class AB amplifier is operating significantly out of class A the positive supply rail ends up being modulated by the positive going half of the signal alone with the reverse happening for the negative. When you decouple the power rails the decoupling capacitors couple some of this into the ground and contaminate it. This is why there are two ground listed in the schematics. The decoupling capacitors create a dirty ground into wherever they are connected and the sensitive areas of the amplifier end up seeing this and it destroys the amplifiers performance.

The way to go about making sure this doesn't happen is to split the grounds up inside the amplifier. Any decoupling capacitor should basically be connected to the same trace on the PCB and then this trace be connected back to the common point back at the reservoir capacitors.

The other areas on the PCB, such as the grounds where the current sources, the feedback network and the input network connect to need to go back to the common point via their own wire, however this in and of itself isn't enough. What you need to do is take a short piece of wire from the common point at the reservoir caps and connect it to a small piece of PCB. You then connect the wire from the feedback ground back to this piece of PCB. This keeps the dirty and clean grounds separate.

The return from the load, ie the negative loudspeaker terminal, needs to connect back to the clean ground and anything else, such as muting circuits needs to go back to the dirty one.

Doing the above sorts out one of the most insidious forms of distortion, the contamination of the clean signal ground from the dirty ground.

The next thing that happens with the modulated signal on the power rails is that it can inductively couple into sensitive areas of the amplifier circuitry too. If you've got power rails coming into the board via flying leads then how they are positioned can be critical. You need to keep them away from the input stage. This distortion gets worse as frequency increases and usually isn't particularly bad at 1kHz. That said, if this distortion is being a problem is can also be due to bad layout and cannot be fixed without a board revision.

If these two things are sorted out then there are a couple of other things that can cause a problem with the amp itself but they are usually quite benign.

The problem when you connect the amplifier to the computer is that you create loops between the mains wiring, the PC, the input connectors of the amplifier, the output of the sound card, the input of the sound card and the output of the amplifier. There is one important thing that you must make sure does not happen with this set up, otherwise it will destroy the measured performance. What can happen is that the return current from the load can end up flowing through the shield of the input connector/cable. If this happens it's bad.

Hopefully some of that might give you some things to try.

Thanks for this analytic help.
I have take care for most of these.
Finally the problem seems to be the negative loudspeaker terminal .This terminal isn't possible to connect to" line in "GND.

 

[thimios]

Hi Thimios,

Yes I can do that. Tomorrow morning sometime. I have a question for you. In your screen shots of the square waves at 100mHz and above, what kind of input cap are you playing through? The only way I have been able to get one to look like that is with the input cap bypassed.

Thanks, Terry

Terry i haven't any extreme part,
I always use cheap electrolytic.
If you can see this now please say the exact post and i will give you a photo

 

[thimios]

What RMS voltage are you measuring at the OP? The problem with sound cards is the absolute level is relative and you need a reference somewhere. I think that hum / buzz would be audible if you placed a speaker across the output (with no signal) . Try disconnecting the mic input gnd connection. This may break the ground loop.

Output= 30V RMS.
I didn't test with microphone.
Amplifier output direct coupled to sound card "line in"via ATTENUATION.
No hum/buzz even your ear close to the speaker

 

[ostripper]

Hi OS,

That did the trick. No more oscillation and playing beautifully.

As for the UV's working as a clipping indicator? You have better have some pretty impressive speakers and a big room. I don't think I could get this to clip with my B1. Bet you need over 2V input to get there.

Hey OS, maybe you should edit post #1 to direct folks to the latest schematic so they won't have to hunt for it.

Oh yeah, how important are the 3K9 resistors? I have the 5K6 per the earlier schematic. Do I need to change those.

Here's a pic of it playing happiliy along.



Blessings, Terry

Cool ... 3K9 is for low rails , your 5k6 should be good for "monster " rails.

You have 90V rails ? .... your headroom must be nearly unlimited. I could
never get my 72V rail "supersym" to ever clip.

90V Clip should be 162V p-p @ 2.9V input. If you want to be outrageous , drop R7/10 to 68R ... stability is the same , but clip = 162V@ 2.2V input.

OS

 

[still4given]

Hi OS,

No I am currently running +-77V rails. The 90V rail OPS is currently waiting until I decide which one will work best for that. It will reside in the case currently occupied by the Leach Superamp. I have already built the OPS boards, just waiting to try some more of these IPS before making a decision. It will be cool if the best turns out to be the CFA-XH since I have extra boards and it is cheap to build.

I already tried to get it to clip with my sine wave generator but it only puts out 1.3V and wouldn't even come close.

 

[PMA]

I already tried to get it to clip with my sine wave generator but it only puts out 1.3V and wouldn't even come close.

You might like to use even the simplest (opamp) preamplifier with gain 5x or 10x to get enough voltage swing behind the generator output?

 

[still4given]

You might like to use even the simplest (opamp) preamplifier with gain 5x or 10x to get enough voltage swing behind the generator output?

Got a link to a circuit for that? I might try building one. I do have some industry preamps. I may see if one on them is clean enough and to serve as a boost. I just didn't want to interject any other signal to skew the results.

 

[PMA]

Any decent opamp, like LME49710, OPA134 etc., in a basic circuit according to its datasheet, will do the job more than well for you.

 

[triplej]

Output= 30V RMS.
I didn't test with microphone.
Amplifier output direct coupled to sound card "line in"via ATTENUATION.
No hum/buzz even your ear close to the speaker

So the signal peak is therefore sitting at +30dBv and 50hz peak at -54dBv and would be audible. Anyhow looks like you've fixed it by the ground lift on the input side. Be interested to see updated plots.

 

[bimo]

So the signal peak is therefore sitting at +30dBv and 50hz peak at -54dBv and would be audible. Anyhow looks like you've fixed it by the ground lift on the input side. Be interested to see updated plots.

Me too.

 

[BV]

Devices saturate and 'starve" the led's - cool !

Look at Q1, Q2 colector current (and so power dissipation) during clipping...Also Q6, Q9 base current.

 

[still4given]

Terry i haven't any extreme part,
I always use cheap electrolytic.
If you can see this now please say the exact post and i will give you a photo

Hi Thimios,

I looked back and I can see that you are using a bipolar 3.3uF. I was just wondering because this CFA-XH calls for a 4,7uf which is what I installed. My 100kHz square waves are slightly more rounded than your 200kHz. They are much less rounded than my VSSA which has a 10uF cap. When I bypass the cap then the roundness goes away. It appears that the big cap is having a big affect on things past the audible range. I just wonder what, if any affect it has on what we actually hear.

I just got in from my workshop where I was able to do an A/B listening test with the PMI VSSA and this CFA-XH. With the volume equalized, the PMI VSSA seems to be just a slight bit "tighter" sounding. It may be that the CFA-XH has a slightly more extended bottom end and that makes it sound somewhat looser. I am being very picky here because my biggest thought listening was , "how is it that I am fortunate enough to own an amplifier that sounds this good?" I mean, these are both absolutely stunning.

The A/B setup is this. Technics CD player into a Mesmerize B1 buffer. From there it goes to a box with dual 15K stereo step-attenuators that feed each amp. The amps then feed into a speaker switch which feeds into one set of speakers. The volume pots are set so each amp is producing equal output power and then the speaker switch toggles between the two amps while they are playing.

For this test, I had the switch feeding into 4 speaker cabinets. A JBL 4425 and JBL 4412 on each side hooked in parallel for a 4ohm load. I do this because the 4425 has a more extended bottom but the 4412 has a better high end. This makes the sound stage huge and also provides a 4ohm load. I can unplug either cabinet to try out other scenarios.

At the end of the test this morning I actually had the VSSA at unity gain with both the B1 and the 15K pot at full volume. The CFA-XH was slightly attenuated to keep the output equal. This was very impressive. I must say, it does pose the question as to whether it is worth investing in multiple outputs and huge transformers when a two output, +-36V amp can basically hang right in there. Maybe it is just testament the greatness of the VSSA design.

Attached is a pic of the speakers. Forgive the mess, I had to move things around to retrieve the Aleph-X chassis and just haven't taken the time to reorganize.

Blessings, Terry