Simple Attenuators - Design And Testing

[Baby Thomas]

Sorry, I forgot to explain about the heat sink. This was the only size I was able to find that is big enough to fit all the resistors and is somewhat cheap at the same time. It's taller than I would want/need, but... this is it. It doesn't get warm at all, even after 2 hours of playing. The amp was with two KT88 power tubes, however the HT voltage is "only" 470vDC or so and the MV is at 30-40%. I didn't measure how much watts it can produce, but I don't think it will be more than ~50W of clean power. The attenuator was used like you see it on the last picture - without case/chassis which helps (I think). I went for the bigger wattage resistors so I can run safely 100W head (if needed) and both the resistor and the heat sink are polished and there's a thermal compound applied between them. If someone doesn't know - the power ratings of these resistors are valid only if they're mounted on a proper heat sink, without one their rating drops to maybe... 30%-40% of the max (mine are cheap Chinese resistors, I don't know if there are valid datasheets). I definitely went crazy with the cooling, but I prefer to have safe margin.

 

[Boysenman]

Hi Again JohnH. I'm sorry for coming up daft, and taking up your time like this. Would it be possible for you to send me the spreadsheet, or in any way point me in the direction of how to calculate the right parameters for the loadbox, for the small 10 inch speaker? When looking at a similar speaker, the eminence 1028K, they use a 6oz plug, meassuring 0.5mH of voice coil inductance. I understand why you thought 1mH was high. The Weber 10a100 uses a 7oz plug, so by a rough estimate, I'd guess 0.6mH would be adequate. The amp is a 5f2a as said, so the output impedance should be fairly high per your statement.

My understanding is that your loadbox design adapts quite well to various systems, and thus I'd assume only slight chances would have to made to approximate a small magnet and speaker cone.

Kind regards
Boysen

 

[JohnH]

hi @Boysenman

I'll have a go at matching to a 10" speaker. (I don't send my spreadsheet out), it will be interesting. I expect that the resistors won't change, and also that even with no change, the tone will still be fine. You'd get the tone of the true 10" speaker out of an amp that thinks it's driving a 12. Probably not far off really.

 

[sympatico]

Hi JohnH and everybody else contributing to this thread.
I´ve built an M2 and I can say it delivers a very (for lack of a better word) "tactile" response, far better than my other attenuator.
A big THANK YOU JohnH for sharing!!!

Now, my build has a strange ringing/oscillation going on that I need to get rid of. It´s that raspy grainy fizzy high frequency artefact riding on top of the guitar sound that we all know and dislike. At first I was suspecting my amp (homegrown ab763 Deluxe Reverb), started messing with tube bias and PI caps and so on. Then I discovered that my other attenuator doesn´t reproduce this ringing. Next I found myself unwinding the M2´s inductor, trying to blame it for misbehaving. All to no avail.

Hopfefully I´ll manage to upload a short mobile phone video of my (cheap) oscilloscope in action of visualising the clipping. First part is without the M2. Amp into dummy load, fed with a sinewave from my (cheap) tube audio generator. Next to some children playing on the street you will hear a bit of the sinewave that is emitted by the whole installation even without a speaker connected. After the cut in the video it´s with the M2 in circuit. I imagine everybody will agree to congruence of visual and auditory phenomena...



My guess is - maybe it´s the very cost-effective (dirt-cheap) switches????
Anyone had a simliar problem?
Any advice and suggestions will be greatly appreciated!

 

[Gene Ballzz]

@sympatico
First, to the forum!

Do you have another amp you can test this with?

Just Askin'
Gene

 

[Trouble Free]

Post a picture of what you have.

I don't think I have that problem with mine.

My signal generator died so when the replacement arrives I can test it on the scope.

Hopefully someone else will beat me to it.

 

[sympatico]

@sympatico
First, to the forum!

Do you have another amp you can test this with?

Just Askin'
Gene

Hi Gene, thanks for the welcome!
Yes, do have and will test with other amps. But my M2 is currently not in a working state - had to cut a wire in order to be able to wind up the inductor again. For now I´ll wait until the new switches arrive, there´s other work to do also...
By the way, I did some tests with just letting the amp rip vs attenuated. Absolutely no pronounced fizz without attenuator. The offending frequency was there but not as pronounced and not appearing in the same manner, meaning if for example you play a chord and let it ring out, the fizz doesn´t ride on top of the notes and doesn´t (hard to put into words: ) have this sputtery flakey trails going on that slowly disappear in a sort of wavy fashion... if that makes any sense...

 

[sympatico]

Post a picture of what you have.

I don't think I have that problem with mine.

My signal generator died so when the replacement arrives I can test it on the scope.

Hopefully someone else will beat me to it.

 

[Trouble Free]

My guess is - maybe it´s the very cost-effective (dirt-cheap) switches????

I am going to guess them switches also, I tried some similar ones when I started mine. one of them fell apart after soldering. I decided on heavy duty ones after that.

 

[sympatico]

one of them fell apart after soldering

ja, had that happening, too. One contact gave up with 3 seconds of 220°C on it. If you look closely, these switches do not have the contacts epoxied (?) in. Since there´s no room for big switches I now ordered Miyama MS series, these are usually very nice! I will post results when it´s done - the order may take 7 days or so to arrive here...

 

[sympatico]

The new switches do have the better contacts:

 

[JohnH]

hi @sympatico and welcome to our thread.

Looking at those traces, it looks like the difference is in cross-over distortion? (or something like it except it only happens at high drive levels?), ie its the small up/down as it crosses from + to - rather than at the peaks? Its there with the dummy load too , and you also note that the frequency is there just with a speaker, but the reactive M2 is enhancing it. If the amp is doing that then maybe the bias of the amp is an area to look at.

Id doubt the issue is with switches, although good switches are important.

From the traces, when the signal gets big enough to clip at the peaks, I'm not seeing anything unexpected.

If you have crossover distortion, then for that instant as it crosses over, the output is uncontrolled and the reactive circuit has free reign to respond. A resistive load wouldn't do this.

But, clearly what you have found from playing indicates a difference in M2 vs a real speaker. There's a couple of things you can try:

First, make sure there is no steel bolt through the coil, that makes a huge difference to the inductance!

Second, with M2 you can simply shunt the coil, ie bypass it with a solid wire and this will leave you with a fully resistive load, but of the right ohms and it should still sound fine. That may bring this added tone back down to where it normally is. (Id still think you want to get rid of this tone at the amp though). If that test worked, you can instead, bypass the coil with another power resistor in parallel with the coil which will just tame it a little at high frequencies. What value for this resistor? If you have an 8 Ohm build using a 0.9mH coil, then I'm thinking about 22, 27 or 33 Ohms to trim its effect above 4 or 5 khz.

Good luck and let us know!

 

[JohnH]

hi @Boysenman

I'll have a go at matching to a 10" speaker. (I don't send my spreadsheet out), it will be interesting. I expect that the resistors won't change, and also that even with no change, the tone will still be fine. You'd get the tone of the true 10" speaker out of an amp that thinks it's driving a 12. Probably not far off really.

:

Hi @Boysenman Here's some stuff relating to using a 10" speaker:

10" Speaker model

This is the model I came up with to match the plot that you posted:



The original plot is at left. I pulled a few key points from that and plotted them, as the pale grey line on the right chart. I base my calcs on 16" ohm speakers, so everything is x2 compared to 8 Ohms.

The yellow values are the model, based on Aiken, and the plot of that is in red, showing a reasonable match. Blue is the measured values for the G12M cab from Mike Lind.

The peak in the 10" plots represented free-air or open-back performance. It would be higher frequency in a 10" closed back system.

M2 adjusted

To match the impedance of the M2 design to this 10" speaker, I dropped the L1 inductor from 0.9mH to 0.6mH. This is the impedance as seen by the amp, in Ohms. Red is the 10" speaker model, Blue is M2 set at -14db:




Output curves

Here's the output, stepping through all the attenuation settings, with red being the full speaker:




So, to match to a 10" speaker, the basic tone output actually doesn't change from the 12" for small signals which is as I expected. But to match the impedance a bit closer for the amp, L1 can reduce to 0.6mH.

 

[sympatico]

Hi JohnH, thanks for the welcome and for your comprehensive input.

- crossover distortion: I already did tests with biasing (I even changed out the coupling caps to the output tubes, thinking maybe one is leaky...) for the same reasons you mentioned. The graps where recorded with a fairly high negative bias voltage resulting in low bias current of around 15ma per tube. Typically these amps run with 20 to 25 or so. High crossover distortion was to be expected. I came to set such low current in a very unscientific way by playing a chord that would provoke lots of the "unpleasant stuff" and dialing in different bias voltages. Low bias current gave less "unpleasant stuff" with different volume settings.

You are right with saying the difference of the two graphs is mainly in the crossover, but there also sems to be more ringing in the compressed peaks. (screenshots attached)
(Whatever that means, I don´t really understand a lot about analysing these graphs really.)

- inductance: I did use a nylon screw with a stainless nut (non magnetic). I did unwind the inductor to see wheter it´s too much. Then I did shunt the coil with an alligator clip. No effect concerning the flaky stuff.

- tone is in the amp: Ja, I mentioned the offending frequency is in the amp even without attenuator, but it´s not behaving the same way. Well... if I think about it again I realise I cannot know that because it is just so much quiter that it maybe even does behave the same way and just gets masked.

Still, my assumption is there must be somthing else giong on.

Things I will reconsider and investigate:
- do messing-around-with-bias-voltage test again and check for crossover distortion specifics.
- do purely resistive attenuator testing
- definitely do test with another amp (!!)

I will report back!

Edit:
- I wonder if perhaps a loosely wound inductor could be a factor??

 

[telesto]

So...with all the riots and Coronavirus going on, I decided to hide inside from the world and do some electro-resistor-therapy Impressed by my build quality, no doubt? I played with some different configurations, and went with a 7dB+13dB+13dB. That gets my 10W and 22W amps down nicely to bedroom-levels. I played for about 5-10 minutes and my a-hole neighbor didn't bang on the wall, so it works good ....or he's not home...one or the other

Anyway, audio quality is MUCHO better than with the L-Pad that I had wired between my amp and speaker I had the volume up in the 7 to 9 area to get the power tubes to breakup, and the big 50W resistor did get pretty warm after 5-10 minutes. Ok, it's not "heat-sinked" to a metal box, I guess that would help instead of my "al fresco" design (please nobody ask who Al Fresco is, I won't respond

Actually, I did buy a metal box to make a final build in. Since I don't need all the stages, just a "one stage" attenuator of ~33dB, I'd like to see if I can reduce the number of resistors needed. My guess right now, is that it probably won't be possible with fewer resistors to keep a ~8ohm load on the amp, and have the speaker still seeing impedance in the ~18ohm area. But I'll play around with some numbers and see what I can come up with. If John or anyone has done some math already and already figured something out for that, please let me know

Big thanks again John, great project!

 

[_Steve]

Love it!

 

[telesto]

Now, my build has a strange ringing/oscillation going on that I need to get rid of. It´s that raspy grainy fizzy high frequency artefact riding on top of the guitar sound that we all know and dislike. At first I was suspecting my amp (homegrown ab763 Deluxe Reverb), started messing with tube bias and PI caps and so on. Then I discovered that my other attenuator doesn´t reproduce this ringing. Next I found myself unwinding the M2´s inductor, trying to blame it for misbehaving. All to no avail.

Any advice and suggestions will be greatly appreciated!

I read/heard somewhere inductive attenuators and amps with NFB/Presence may have oscillating issues. Does your amp have NFB? Or else maybe try to take the inductor out of the attenuator and see how it sounds just with resistive components?

 

[JohnH]

Hi @telesto
if you just want a fixed attenuation with no switching, then with the 8 ohm and 18 ohm values as a target, you can make any attenuation you want as a single stage so long as its more than 7db. Youd use 3 resistors in a T or Pi arrangement, and they can be direct equivalents to each other. So if you are just wanting a resistive version, its just three resistors. If you want the inductive stage 1, Then you can pick that from design M or M2 and then have three resistors for the rest. The two-resistor stages used in designs so far are a simplification, but work fine when hidden behind Stage 1 and within the range 3.5 to 14.

But are you going for -23 or -33 db total?

 

[telesto]

Hi John, from testing today, -33dB was the good value for me, so I'm shooting for that, with no inductor (oops, I put 23dB in the earlier post, that was I mistake, I corrected it now). Yea, I was also starting to look at a T-attenuator. An arrangement of 7ohm/1ohm/7ohm looks interesting, I didn't work out the dB yet, but the calculation with Watts I made would bring a 10W amp to about 0.01W...oh wait, I guess that would be 30dB now that I look at it (oops, no it's 24dB according to the T-calculator) What's also interesting is that with the 1ohm shunt resistor, I can put any other resistor in series with the speaker to reduce more power, and it has no affect on the impedance the amp sees.

...oh, I forgot to do anything with the inductor. I do have some now (0.8mH, 0.39mH and 0.56mH). I wanted to work out the volume level first with the resistors. Maybe if there's more riots next week, I'll stay home and play with the inductors

 

[JohnH]

Cool, so I can do that in a spreadsheet. All it is is a few simple series/parallel calcs, then using 'goal seek' to home in on a solution. Since we want three things being two impedance results and an attenuation, you goal-seek for one result at a time, varying one resistor for each parsmeter, and keep cycling it and it all converges. Put all that as a macro and link it to a button and it's a fun high-speed auto attenuator designer!

Anyway, for -33 db, 8 ohm in and 18ohm out, and assuming the amp is effectively also 18 ohm:

As a 3 part T attenuator:
Amp 18 ohm equiv
First series resistor 7.4
Shunt resistor 0.58
Second series resistor 17.4
Speaker 8 ohm

To tweak that to nearest standard values, it's 8, 0.56, 18 .

Let's say you have stage 1 at -7db to get the inductor Then you want another -26 db, you need: 6.8, 1.3 and 16.8. Nearest standard values 6.8, 1.5, 18.

Both of those theoretical T designs have exact equivalent 'Pi' designs too.( ie, shunt, series, shunt)
-33db 8.3, 247, 19.4
-26db 8.6, 110, 21.3

Again, nearest standard values gets you close.

So have fun rummaging through the spares box!