|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
40.136.248.102
Went through the first two videos that Cougar posted the links to a couple of times and was then able to get the attached traces at two different time rates (see, already using the lingo). Seeing how to do it in the videos really helps! The leads are attached to the output trans output leads with the speaker attached as well. Shorting plug is in the input rca of the amp. This isn't quite what I expected to see. I thought it would be a nice sine wave with the some junk on it from ac. What am I seeing here? Looks like some superimposed sine waves plus some spikes. Could the spikes be the ripple? Maybe the 3rd video tells how to interpret what you get on the scope.
Edits: 10/24/16 10/24/16 10/24/16 10/24/16 10/24/16Follow Ups:
Attached are the schems for the 26/813/813 amp, the psu for both the 813 and the 845 (it's the same for both amps), and the 10vdc for both amps. To answer Russ's question about the caps, as shown in the schem, there are two sets of caps. Each set consists of a old stock 40 uf oil cap and a new stock 47 uf solen tied in parallel for a total of 87 uf per each set, Each cap has a 7k bleeder resistor across the terminals.
Edits: 10/25/16 10/25/16 10/25/16 10/25/16 10/25/16
The four 7k resistors are drawing 271ma. of current from the power supply.They are not needed.
Figure out what the total current draw of the circuit is (all the tubes feed by the supply) and then make sure the input choke has enough inductance to be critical.
Tell me the total current draw and I tell you how much inductance you need.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 10/25/16
I sort of figured it was wrong from what I saw cruising around on the web tonight, but I could not find anything that directly addressed what I was trying to do. All filaments are supplied from separate transformers, so the total current draw would be 2 x 62mA for the 813's plate current and 6 mA for the 26 plate current (only one channel right now.) for 130mA total , yes? Also, I think I only need one resistor on the first cap and that's all right? Thanks!
Edits: 10/25/16 10/25/16
"130mA total "
Yes.
The math says 3.75Hy is critical for 475 volts at 130ma. so I would use 5Hy or maybe up to 7.5Hy.
"Also, I think I only need one resistor on the first cap and that's all right?"
With a 5Hy choke and 130ma. current draw you don't need a shunt resistor at all.
If you wanted a bleed resistor just so the cap voltage will bleed down when you turn the thing off, for safety, then a 220k would draw 2ma. and dissipate 1 watt (use a 3watt).
What I saying is this, if your choke is critical inductance, with the current draw of the circuit alone, then you don't need shunt resistors to increase the current draw.
Adding extra current draw will make the filter less effective in terms of ripple.
BTW "balancing resistors" are when you have two cap in series and you want to keep the voltage balanced across each cap. Each cap, in the series, would have a resistor across it. But never as small a 7k. 7K draws a lot of current. Those balancing resistors would be more like 100k.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
What I saying is this, if your choke is critical inductance, with the current draw of the circuit alone, then you don't need shunt resistors to increase the current draw.
While this is true, it isn't the whole story particularly if using a solid state or directly heated rectifiers. In this case the B+ comes up much quicker than the tubes and without the tube current draw the B+ shoots up to the full cap input value for a moment. Sometimes this doesn't matter but consider a 400V choke input supply and some 300B's heated with coleman regs. Before the output tubes are hot, the B+ will jump up to over 600V which isn't too big a deal for the tubes since current draw will be 0 but it might be a bit of a surprise to the caps (both PS and coupling)
I have an amp here that uses damper diodes for rectification and coleman regs for the 300B's and they track each other perfectly on initial turnon but if you shut the amp off for a moment and then turn it back on, the dampers conduct almost immediately since they are still warm but the filament regs take the normal time to gently ramp up causing a temporary overvoltage situation that needs to be accounted for and I have some shorted 600V coupling caps to prove it.
dave
I have separate heater and filament transformers that come on with the main switch and after a minute I throw the switch for the plate transformer.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
To do the heater and B+ transformers. I would like to do this on my Aikido preamp or implement a timer circuit like the one counterpoint uses in their upper preamps and power amps.
Do you use a NTC like an CL90 inline on the switch side for the main transformer?
No I don't, hasn't been a problem.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
That's my standard method too, unless I am using a slow turn on tube rectifier like a 5AR4.
Sounds like my 5-25H swinging choke ought to do the trick. I will put a bleeder resistor on the last cap then. Also, when I get the other channel set up is it correct to assume the critical inductance becomes 7.5 Hy? Is there a link to the equations that covers the calculations for this? Thanks for the help!
Edits: 10/26/16 10/26/16
chapter 14
Thanks. Reading that ought to keep me out of trouble and off the streets for a while!
The formula for critical inductance is,
inductance = ((voltage/current) + dcr of choke) / 1000
Most books (including the one linked by elblanco) state that one should use 2 times the calculated critical inductance.
"..when I get the other channel set up is it correct to assume the critical inductance becomes 7.5 Hy?"
No, if you draw twice the current at the same voltage then critical inductance will be one half the value originally calculated but you have to figure in the DCR of the choke so it won't be exactly one half.
The math at 130ma. current draw
475/.13=3655+100 (drc of choke just a guess)=3755/1000=3.755Hy So twice that would be 7.51Hy
The math at 260ma. current draw
475/.26=1827+100=1927/1000=1.927Hy So twice that would be 3.845Hy
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
That's weird and counter intuitive, so it is probably true. Thanks, I will make sure I remember that when the time comes. Two more sheets for the notebook ......
Tried putting a swinging choke 5-25H in place of the first 8H choke in my psu and the ripple dropped from 468 mV to 210 mV. Still too much but a big improvement. Now I am trying to figure out why that happened and how I can apply that to the other choke and try to lower it some more. I don't have another swinging choke unfortunately. Is there a way to measure what value a swinging choke is at when the psu is powered up?
"Is there a way to measure what value a swinging choke is at when the psu is powered up?"
Yes, see my posts in the link below.
When you say the ripple dropped, where are you measuring this in the circuit?
--------------------------
Buy Chinese. Bury freedom.
Yes, I attached the dvm probes to last cap. Also, I just got through adding a 7k resistor across each cap to balance the load and help with limiting in rush. Total of 4 caps, 2 groups of 2 in parallel for 87uf in each group. When I checked the ripple this time (same way) it had dropped to 45 mV. Not sure why that happened, but would like to know. Also, I am wondering if I raise or lower the values of the resistors if that will drop the ripple even more. Thanks for the link, will check it out.
Edits: 10/25/16
If the addition of bleeder resistors causes ripple to go down, that might imply the amplifier itself is drawing insufficient current for the chokes to be effective. You can model that in PSUD. 45mV ripple at the output stage probably equates to around 2mV at the speaker. Have you posted a complete schematic of the supply here?
--------------------------
Buy Chinese. Bury freedom.
If the addition of bleeder resistors causes ripple to go down, that might imply the amplifier itself is drawing insufficient current for the chokes to be effective.
are you referring to AC or DC current?
can you give a scenario where this might happen?
dave
If the inductance of the input choke is less than critical and then we draw more current causing the inductance to be critical the voltage will go down but drawing more current will always make the ripple go up.
Did I get that right?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
but the issue was bleeders causing the ripple to go down. I cannot see anything related to the choke to cause this since increasing current draw always reduces inductance which should make ripple go up.
dave
"but drawing more current will always make the ripple go up."
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
which puts us back to the beginning and me asking why bleeding more current causes less ripple. TK said it could be because the amp may be "drawing insufficient current for the chokes to be effective"
I want to know how drawing current makes a choke more effective.
dave
"I want to know how drawing current makes a choke more effective. "
It doesn't in terms of ripple, it does in terms of voltage regulation.
That is to say, if the inductance is less than critical at the lower current draw and at or beyond critical at the higher current draw.
If the inductance is "on the edge" of being a value that is critical and, with a changing current draw, is falling in and out of critical the voltage regulation will be........bad.
All of that aside, I agree with you. For a given inductance (and cap values) more current draw = more ripple.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Not in it's current form, but I will. I need to do it for my records anyway. I'll try to get to it today. I can hear that the hum is down, so I think I will take a look at it at the speaker again on the scope and see what we get.
I did it in psud, but I did not see anything that indicated a problem like that.
Are they new? What brand and type? They aren't warm are they?
Might be worth throwing a high voltage film cap across the first cap bank...1uF 630 volt should do but the bigger the better.
Could be your caps just aren't up to snuff.
nt
I'd suggest clip leading RCA shell, negative side of speaker connection, power cord ground, and the ground lead of probe all together. You need them to all be at the same potential. Do all of this with the amp off.
With a RCA shorting plug in place I don't think you will see so much.....partly depending on if it is a push pull or SET amp.
If I'm looking to measure ripple voltage I do it at the power supply caps. AC coupling only....auto trigger works okay to start with.
Clip leading things together did not make a difference. I think everything is grounded together as it is. Your comment about checking the ripple at the caps did make me slap my forehead though, as you are right. Put my fluke on the last cap with it set on ac and got roughly o.5 vac, which explains the ripple being so high at output of the amp. Now I just have to figure out why I have so much more ripple than psud2 says I will have. I must have some parameters wrong for the trans or the caps. It usually gives a closer estimate than that.
I will give that a try. I use psud2 to check for ripple before I put the psu together and it usually is pretty close to right on, but I will check it with my fluke and see.
I think I may have found at least part of the problem. I decided to switch the probe to 1x instead of 10x to see what would happen,and when I did I noticed that my scope has little brackets marked 1x and 10x under the volts/division dial on each channel. When I told TK the volts per division was 0.1 I was reading a number in the 10x window because I had the probe set on 10x. HOWEVER, if you switch the probe to 1x and get the same size trace, the 0.1 is in the 1x window instead. In other words, you don't have to correct for the 10x probe if you read it at the right place on the dial. Soooooooo, that means that I am actually seeing 2.5 divisions times 0.1 for a total of 0.25 volts total instead of 2.5. Still not the best, but much better. It is about 100 times higher than what psud2 predicts I should be seeing with my psu though, so I am wondering if I am missing something else in all this or if I am picking up a lot of stray ac somewhere else along the way.
Edits: 10/24/16 10/25/16
nada aqui
--------------------------
Buy Chinese. Bury freedom.
Be sure the neg of the scope is on chassis ground. What is the amplitude/frequency of that signal? If you're not sure, just tell us the settings for V/div and sweep time/div, and whether that's a 10:1 probe.
--------------------------
Buy Chinese. Bury freedom.
The negative clamp for the probe is connected to the ground of the rca cord from the amp, so that should be chassis ground. The v/div is the same for both pictures and is 0.1 v/div. The time sweep is 2 ms for the first trace and 5 ms for the second. Probe is set at 10:1.
Looking at the last trace, that works out to approximately 2.5V p-p amplitude (.1V/div X 2.5 divisions X x10 probe). The large peaks in the last trace are about 3.25 divisions apart horizontally, or 16.25 mS. The inverse of that is ~ 60 Hz. There's also considerable 120 Hz energy. I don't see anything else happening, such as oscillations.
--------------------------
Buy Chinese. Bury freedom.
2.5V P-P at the output terminals with the speaker attached?
So that means I have way to much ac getting through the filters of the psu, right?
What part of the picture shows the 120 hz? Could it be the 120 hz that is causing the small buzz I hear?
Vinnie, the 120 Hz energy is visible in the form of smaller peaks that fall in-between the 60 Hz energy. They're also occurring "underneath" the 60 Hz peaks, but being masked. This could all be caused by a number of things, and maybe more than one. Generally speaking, 60 Hz can be picked up from the PS magnetics or from the surrounding environment. 120 Hz might be from the PS filter or filament doubling (my 211s did that). If you want to probe the B+, you'll need a high-voltage probe. However, a simpler method is to model the PS in either PSUD or SPICE. If you know all the characteristics of the components in the power supply, the simulators will be very close to real world. 2.5V is quite a lot of ripple, so I suspect this is occurring in an early stage and being amplified.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/24/16
If it isn't in the psu, it might be coming from the rats nest huh? I may have to wait until I clean things up with the final build before I can get rid of it.
I did model the ps in psud2 before I put it together and I just did it again after double checking the schematic against what is out there. I have a 1500 vct trans with 3 1N4007 diodes in series off each leg going to the first 8H choke. There is an 87 uf cap between that and the next 8H choke and then another 87 uf cap with a 20 ohm/50w resistor across it. When I run this in psud2 I get what looks to me like only about .003 volts (3mV right?). That should be ok shouldn't it?
Edits: 10/24/16
vinnie, without changing anything else, move the switch under the volts/div knob for channel 1 to AC and snap a photo.
It was already in the ac position. I moved it to the ground position and it changed to a flat line, and then I moved it to dc and got exactly the same trace as in the ac position. I then switched the probe to channel 2 and got exactly the same results. I have a feeling you guys are going to tell me that is not what should have happened. : )
Right - the output should be AC only. If you are in GND, then you just get a flat line: this position is there so you can use your vertical adjust to put that one one reference grid. If you had then, let's say 2.5v DC AND your AC signal, putting on DC would make the trace shift up by one division (2.5v) - essentially a combined AC and DC volt meter.
If you have the input shorted then your original posted wave is the results of all the odd artifacts added by your amp because, what I don't think has been posted, is that when the input is shorted, the output should (ideally) be a flat line: no AC, no DC. I agree with someone who suggested this is from the heater arrangement you have.
In a DHSET, you will likely NEVER have a completely flat line, but should be closer than what you have. If you move the time base to a faster and faster time, you may see those spikes become their own little sign waves on top of the dominant sign wave - an oscillation.
By the way, I don't know where you live, but I'm in the Portland, OR area and have the same scope. Happy to walk you through in more detail.
'thanks for the post. I am in NC, so there is a bit of country between us. The videos are helping a lot, so I hope to get the hang of it. Not sure what you mean by the heater arrangement I have; is there a better way to do it?
He doesn't need to do that. There's no DC on the secondary of the OPT.
--------------------------
Buy Chinese. Bury freedom.
I thought he was on the other side. never mind.
Then how come it shows an ac curve in the DC position on both channels?
Vinnie, the scope merely inserts a series blocking cap in AC mode. If there's no DC on the point you're analyzing, moving the switch from AC to DC won't make any difference.Edit: The results you reported above are exactly what should happen.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/24/16 10/24/16
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: