[runeight]

Say, gents, I have been on the road for a bit, but I see that my comment about the transformers didn't reflect as much humor as intended. Sorry about that.
 
I think we may be over analyzing this problem. The voltages you guys are showing are probably alright. To me adding CRC circuits, etc. is adding more complexity than needed (but thanks for the help Marc

 ).
 
38-39V on the regs is not a serious problem, just higher than designed.
 
In the future my guess is that sachu's idea of using the avel 28-0-28 transformer might be the perfect choice or the sumr with a 28-0-28 secondary.

 

[fishski13]

Quote:

Say, gents, I have been on the road for a bit, but I see that my comment about the transformers didn't reflect as much humor as intended. Sorry about that.
 
I think we may be over analyzing this problem. The voltages you guys are showing are probably alright. To me adding CRC circuits, etc. is adding more complexity than needed (but thanks for the help Marc

 ).
 
38-39V on the regs is not a serious problem, just higher than designed.
 
In the future my guess is that sachu's idea of using the avel 28-0-28 transformer might be the perfect choice or the sumr with a 28-0-28 secondary.

CL-60 thermistors for power-on voltage transients?  larger heatsinks?

 

[qusp]

guys, just as reassurance, when testing out higher voltages on a modified D1 mosfet IV, I have run the LM317/337 at +/-65-67v in and +/-55v out. as Runeight has said, its really only the input/output that matters with these floating regs. they do not care what the voltage is, they only care what they are expected to swallow. you could probably run them at 500v out if the V in was lower than 540v

 

[stixx]

 
 
 
Quote:

they do not care what the voltage is, they only care what they are expected to swallow. you could probably run them at 500v out if the V in was lower than 540v

Yep. See them being used in a Maida reg...

 

[ujamerstand]

So as long as the input voltage ramp up slowly so that the regulator don't instantly see the full 40+ volts on startup, we're fine I guess?

 

[jdkJake]

Nice when a plan comes together, even when it is not planned!
 
Had some time this morning before the game (and during the game once it turned ugly). So, I built and tested a couple of these bad boys:
 

 
Very nice. Both worked great until I shorted one across U1 6/7 (can you say visible spark!) while performing the setup. Needless to say, torched a transistor with that graceful move. In any case, repairs were made and much more care taken during setup and I was back in business (U1 pin 6 needs a TP for the careless like me on a Sunday afternoon).
 
In any case, these work great. I socketed R11 and bumped it to 1M, but as I learned, it really does not matter. Since these track the DC offset, they always "do the right thing". With no load, the 1M gave about 6 or 7 seconds of delay. In circuit, upon actual power on, it turns out to be around 20 or 25 seconds as things settle out. Since we are using two, each side come to the party at a different time (usually a couple seconds apart). Kinda weird, but, it works, so, who's complaining. In any case, the stock build (R11 == 475K) should be fine.
 
Of course everyone wants to know, what's the impact on power?
 
MOSFETS at ~220mV
AC at ~122.5VAC (was around 120 earlier today, did I mention the joys of Florida Power?)
AC on the secondaries: ~30.6VAC
At the regulator input: ~36.7V
 
Nice. Moving right into the sweet spot. I can live with this.
 
Quick road trip tomorrow (damn day job) and them we case this puppy up and call it day.

 

[runeight]

Gents, it seems like there are 3-4 working amps. Are there any listening impressions beyond jdk's. The designer always wants to know how the design is doing.

 

[cobaltmute]

Well Richard has stated that my transformer should ship by Thursday at the latest.  So hopefully by the end of the weekend I'll have comments.

 

[ujamerstand]

Well, I really suck at listening impressions, but I'll give it a try. IMO, the combo between EHHA and the HE-5 provides more impact than 404s and the exstata. I've never found the 404s to be lacking in that department, but the EHHA and the HE-5 has that oomph that makes certain music that much enjoyable to listen to. The sibilance on the HE-5 is at first unbearable, but I don't think this is a problem with the amp, and I got used to it pretty quickly. The amp is also quite speedy. Probably not as fast as the exstata+404 combo, but the HE-5 running on the EHHA is no sleuth either. I really have to clean up the wiring and case it up to reduce the occasional hum; but overall, the amp is very quiet.

 

[Cloud]

was going through EHHA build thread, looking at information for a planned 3 channel active ground build, thinking that a unity gain is required.
 
then i saw this interesting post by runeight
 
http://www.head-fi.org/forum/thread/422429/cavalli-ehha-embedded-hybrid-headphone-amp/360#post_5733197
 
"There is no need to operate the amp in unity gain mode even for the three channel configuration. The ground channel can have its input grounded and then use the same component values as the other two channels. This will ensure that the ground channel can operate at the same speed as the amplifying channels. The amp is unity gain stable with reduced bandwidth as you can see by the very large Cdom caps that Snoopy needed to stablize the SQ response. I think it's better to give the ground channel the same bandwidth as the amp channels."
 
i suppose this applies to EHHA Rev A as well?
 
can I know what are the OL gain and CL gain that the beta team are using?

 

[jdkJake]

Quote:

 
can I know what are the OL gain and CL gain that the beta team are using?

So, according to Alex, the 75k resistors on R13/14 provide an OL gain of about 500 with a NFB of ~35db. Using the default 1.2k on R11 should provide a CL of around 9.3.  This should yield a total gain of around 9.2 with 35db NFB.
 
For my build, I currently have a 1.3k in at R11 for a CL of around 8.7 (total gain of about 8.5 with around 35db NFB). While I think the gain is probably okay for my HD650's, I think it is a bit too much for the Grados, so I am going to socket the resistors and lower the CL a bit. I may also play a bit with the NFB, but, I am pretty happy with the overall sound and might leave that alone.
 
I am thinking of using a 2.4k at R11 to get a CL of about 5.2. This should yield a total gain of about 5.1 with 35db of NFB. I am thinking this will be about right. Again, I can play with that a bit as the board will be socketed to allow easy change. 
 
Not sure what the others are using.
 
 

 

[cobaltmute]

My intent when I get my transformer, at least for initial try, is to use 10k resistors to set the open loop gain at ~100.  CL gain for initial try will be the default as per the BOM at ~9.3.

 

[jdkJake]

Quote:

My intent when I get my transformer, at least for initial try, is to use 10k resistors to set the open loop gain at ~100.  CL gain for initial try will be the default as per the BOM at ~9.3.

Are you trying more to eliminate NFB rather than reduce overall gain? By my calculations, those values should take you to about 8.5 overall but significantly reduce the NFB.

 

[cobaltmute]

Yup.  the goal is low NFB.
 
I'll play with the CL gain as well, but want to see how the amp sounds to start with.  I'm not a volume hound, so I may do lower gain to get better adjustment on the pot, we'll have to see.
 
Looks like SumR didn't get my transformer out on time as I didn't get a shipment notice., so it doesn't look like I'll be up and running this weekend.

 

[jdkJake]

Funny, I am probably going to go the opposite way and slowly lower the NFB and raise the CL as I go. I kinda want to shoot for a gain of 5 throughout the whole process. We will see though. I am torn through the promise of wide bandwidth and the seduction of low NFB to bring out the tube.

In preparation for that experiment, I installed the resistor sockets for R11,13 and 14 today. Man, what an ordeal. Things are quite tight with everything else installed. New and upgraded solder pump got a real workout as well. Nevertheless, I got it done. I would certainly advise others that, if you think you want to play with the gain at all, socket early!

I'll get a good listen later tonight and see how she sounds provided I didn't mess anything up through the install effort.