[runeight]

I think we should clarify the regulator discussion a bit.
 
The max input/output differential on the LM317 is 40V on the National spec. Since we are operating at 30V at the output there is plenty of room.
 
But there is another reference point, the ADJ pin. The ADJ pin is sitting very close to 30V so that the voltage difference between any two pins on the device never exceeds 40V, in fact given the measurements made so far under load, doesn't ever exceed 10V.
 
Only these relative voltages matter. There are no absolute voltages on these devices.
 
However, these conditions only apply once the regulators are at the operating point.
 
The real problem is on power up. If everything is discharged then the Out and Adj pins of the regs are at 0VDC. The IN swings up quickly possibly to over 40V because the loading takes some time to stabilize. The Out and Adj follow depending on how they are loaded. As long as the relative difference of 40V is not exceeded then the regulator survives.
 
We should add this means 40V in the forward direction. The regs have diodes to protect against reverse biasing.
 
It appears that with these transformers and amplifier start up the Out and Adj are following fast enough so that the max differential is not exceeded..
 
But, as noted, to be safer you can use the HV devices. 
 

 

[jdkJake]

So the main advantage of the HVT parts is the 60V differential tolerance?

 

[ujamerstand]

Right, since the regulators did not blow up on me on the first start up, adj and out are definitely following fast enough. But as a hypothetical situation, where LM337 is knocked out, what would happen to the amplifier? Also, the bias of the amplifier changes as the amp heats up. My pre-reg voltage drops as the bias increases and stabilizes below 40V when the amp finishes warming up. So Vi would be higher than 40V at a cold startup. Are there any situations where the Vout and Vadj are not able to follow Vin fast enough, and the max differential is exceeded? 

 

[runeight]

Yes.
 
Edit: I should always check for new posts!! The yes is in response to jdk's question about the 60V differential.
 
There might be conditions where these voltages could be exceeded but they wouldn't be unique to this power supply or amp. As I look at the circuit I don't see a condition where this could happen, but start up/shut down conditions can be tricky.
 
Depends on what you mean by knocked out.
 
If it is shorted IN to OUT then the negative rail voltage will go to whatever the input to regulator is (approximately). The 79L15 will probably exceed its maximum differential and die. The opamp will have only one rail and will stop working and the output will skew to have a high DC.
 
If it opens, then the bottom rail will float upwards, all differential voltages in the amp will get smaller, and the O/P DC will skew to the positve rail.
 
Gents this is not the only split supply that runs at +/-30V using IC regulators. It is a standard regulator configuration.
 
We cannot guarantee that there will never, ever be a failure, we can just try to keep everything within acceptable envelopes. Many folks who slap together linear regs into a supply like this may not even realize the dynamics of what happens when you turn the bugger on.

 

[runeight]

Having said all of the above, my target was closer to 35-36V as input to regulator. This gives a bit more margin all around. Line voltages are difficult to account for, but the SUMR transformer could be spec'd down just a little bit.

 

[jdkJake]

Okay, so looking back a bit, I found the following:
 
 
   Stringgz is using a 30-0-30 .5A Avel Lindberg and is sitting at 38.2V (unknown line VAC) 
 
   Studeb using a 32-0-32 .8A Antek and is sitting at 39.6V (121.5VAC)
 
   jdkJake was using a 32-0-32 .4A Antek and was sitting ~34V (~121-122VAC)
 
   Ujamerstand and jdkJake are all using the sumR transformer (30-0-30 .4A) and are sitting between ~38V and 39V.
 
Looks like sachu is going to provide a most interesting contrast. 
 
 
BTW, if Stringgz and Studeb can chime in on what version of the LM317/337 they are using, I think that would be helpful as well.

 

[jdkJake]

Also, those HVT parts might be hard to come by, especially the LM337.
 
Digikey has some of the LM317HVT, but none of the LM337HVT. Mouser is out and marks them as obsolete.
 
Might take a bit to locate a source.

 

[ujamerstand]

Thanks for the explanation.

I think I'll keep a few extra components around, and slap on some offset protection on the output of the amp. 
 
jdkJake: What's your pre-reg voltage at startup? Also, I remember you posted numbers for dc offsets, were they measured during cold startup?

 

[jdkJake]

Not exactly sure on the pre-regs at start up, but I seem to remember somewhere south of 38V (~36V seems to come to mind, but, I am not sure). I will remeasure tomorrow. RIght now, the amp is hot and I am tired.
 
Yes, I measured all DC offset from cold starts. I performed that particular measure many times over a couple of days.
 
 

 

[runeight]

And you may wonder why 35-36V.
 
Let's say you're running with 120VAC line and it drops 5V to 115VAC. Your secondary will drop by 1/4th of that approximately (120:30). let's say 1.25V. The peak DC will drop by 1.4X this or about 1.8V. This will track approximately to the DC input to reg, If you're running at 35V this will drop to 33.2V. This is more than enough to keep the reg from dropping out.
 
But remember too that the input to reg voltage is not a flat DC. It has AC ripple on it so even if the average DC is 33.2, the actual voltage is swinging up and down, maybe a half volt or so peak to peak. That's another 250mV drop on the negative swing bringing the instantaneous voltage down to ~33V, bringing it closer to the 30V regulated output. 3V is enough to prevent dropout, but for margin it's seems best not to go lower than this.
 
Thus, getting this right means having the op point so that it is not too low such that drops in line voltage will cause the regs to drop out and such that higher line voltages will not exceed various maximums.
 
And this is why the part of this that I hate the most is when you guys start talking about using every transformer in the book. It's almost impossible to make the design able to handle a very wide variety of transformers, particularly if they have widely varying regulation and secondary DCR.

 

[jdkJake]

Thanks for sharing the design insight Alex. Things are much clearer to me now regarding the intended design margins for the Rail PS.
 
I also understand your frustration with all the part swaps going on. Must be hard to see your design interpreted in so many different ways beyond your control. Kind of like setting your children out into the cold, harsh world. At the end of the day, the implementer is still ultimately the responsible party for a functional (and safe) product that conforms to the design specifications. Please continue to keep us on course, but, realize deviations are inherent in the process due to cost, parts availability and personal preference.
 
I for one, am still happy with the sumR as currently speced. Remember, I am not done loading it down yet. I still have two epsilon12's to hang off of it. While the draw of those parts is not too significant, they do add additional load to the Rail supplies. Hopefully, I should have those built up later this weekend and will have a better idea of how they effect overall performance of the power subsystem.
 
Again, please continue to share your design insights and goals for this amp. It makes the entire project far more personal and a very enjoyable and rewarding experience.
 
We will try not to go too far into the weeds. 

 

[luvdunhill]

Quote:

And this is why the part of this that I hate the most is when you guys start talking about using every transformer in the book. It's almost impossible to make the design able to handle a very wide variety of transformers, particularly if they have widely varying regulation and secondary DCR.

perhaps putting a CRC on board would help with this... along with clear instructions on how to "set the resistor" it would allow for lots of variance in mains voltage, transformers, etc.

 

[sachu]

well i fired up the heater supply. It slipped my mind that the hammond transformers are rated for 115V primaries. Line AC is at 122.4V.
 
Regulator input is seeing 21VDC and am getting 12.6V on the output.
 
The HV transformers are again rated for 115V primaries and secondaries read 32.5VAC. This isn't what i was hoping for as again i thought the primary was 120V windings. Oh well, suffice it to say i will see over 40V on the regultor inputs with the PSU unloaded. Not a good thing IMO. I should have just stuck with my usual Avel dual 28V secondaries transformer. Ah well..we'll see how it does when i bring up the power supplies. Increasing rails PSU caps to 6800uF would help some in dropping the input voltage to the transformers.

 

[stringgz301]

Line voltage is 120.8.  Am using STMicro LM317T for VR+ and On Semi LM337TG for VR-.
 
Quote:

Okay, so looking back a bit, I found the following:
 
 
   Stringgz is using a 30-0-30 .5A Avel Lindberg and is sitting at 38.2V (unknown line VAC) 
 
   Studeb using a 32-0-32 .8A Antek and is sitting at 39.6V (121.5VAC)
 
   jdkJake was using a 32-0-32 .4A Antek and was sitting ~34V (~121-122VAC)
 
   Ujamerstand and jdkJake are all using the sumR transformer (30-0-30 .4A) and are sitting between ~38V and 39V.
 
Looks like sachu is going to provide a most interesting contrast. 
 
 
BTW, if Stringgz and Studeb can chime in on what version of the LM317/337 they are using, I think that would be helpful as well.

 

[TimJo]

Quote:

It slipped my mind that the hammond transformers are rated for 115V primaries. Line AC is at 122.4V.
 

When I buy Hammond transformers, I get the 300 series with the universal primaries. That way you get 100, 110, and 120 taps. It costs a bit more, but it works well here in the NW, and will keep working if one day I find myself living in other places, which is always a dream.