Simple Attenuators - Design And Testing

[JohnH]

@JohnH, many thanks for putting this together! I have a couple of questions:

1. For a 100W amp using the 16 Ohm tap, I presume I need to double the power rating of each component yet again, is that correct?
2. Is the impedance curve (seen by the amp) in the graph only modelled in Spice or also measured, e.g. with the Dayton Audio DATS?

Hi dudu, thanks for your interest.

My numbers posted on this thread are for use with 50W amps and allow a factor of 1.5 for overdriving above 50W, then another factor of x2. The power ratings don't depend on the Ohms value, so double the ratings I list for use with 100W amps, either 8 or 16 Ohms. So for example, R1 gets uprated to 200W, which could be made from 2x 100W resistors, or as matttorando has done, one huge resistor.

I don't have any lab test gear other than a meter. My charts used to develop and present the design are calculated using Spice, or spreadsheets that I have written based on the same theory (and checked against Spice). There are also real frequency response tests which are miced recordings, then analysed using Audacity.

But, to set up my analysis, I am using a real measured impedance curve as a reference, that came from Mike Lind on the TGF. Its based on impedance and frequency measurements from a real 4x12 greenback cab. I set up an equivalent circuit to match it very closely, then use that in my calcs to represent the speaker and to compare with.

 

[yogi.1026]

Hi JohnH and MF'ers, After following this thread closely I decided to bite the bullet and build this design. I stole Gene Ballz idea and went with a 50w/16 Ohm, three 7Db stages with bypass. I have a DSL 50 and 2266 Vintage Modern, with a 1960ax greenback cab, I was fortunate as this design was made with these amps in mind. I ordered all the parts from china, as I live in Western Australia, (3,500km from anywhere) and local stores only stock the same anyway. Everything was at my door within 3 weeks! After reading Genes post on temps I got a little worried, I did some reading of spec sheets and found that the 100w resistors unmounted are only rated at 40w, they dont rate at 100 until they are on a heatsink with 291sq/in.(12x12x1/4in) With this in mind I though I would seperate R1 and R2 onto seperate 3mm plates with HD heatsinks (from a old car amp)mounted underneath. I thought this would help even if techicnaly the only increased the rating by a couple of watts, I used a 200x145x55mm aluminium box and drilled top and bottom for some convection airflow. The 2266 amp LDR vol 8 for extended sessions only get warm to the touch at the resistor, the outside box is cool. Probably way overkill, but I'd rather be safe than sorry. I originally ordered 25amp toggle switches but in testing before use I was getting weird readings, I tracked this down to the DPDT bypass switch. One side wasn't contacting properly, this was at the time John reported that he broke a toggle by dropping it so I went with the rockers,(white were on sale 50% off) bit more involved than drilling for toggles, but hopefully bypass that problem if you are moving gear around all the time. Hows it sound? I was nervous at first(actually googled witch transformer was cheaper)so I went with the DSL. I was always happy with the sound of this amp, It was a case of going from vol1.5/2 to around 4,5/5. this filled the sound out ,less fizzy, more sustain, improvement but not outstanding, moving to the 2266 wow! this is more like it. Bear in mind I'm not going for super low bedroom levels, just to get some power tube goodness, If anyone is one the fence building one just do it. Parts are reasonably cheap, all parts minus the inductors were under US40$. I had to source my inductors locally( many thanks John), still worked out cheaper than China. The only caveat is the actual time to build, I made mine harder by fabricating the heatsinks salvaged from an existing amp, If I did this again I would use the smaller heatsinks mounted under plates with standoffs to secure to the box. Sorry for dribbling so much in a first post but I use this forum as a fantastic resource and would like at contribute something back FWIW. Many thanks to JohnH and contributors to the thread for all the different ideas. Some photos attached.

 

[JohnH]

Hi @yogi.1026 Thanks for posting, and thats a very nice build! Shame to put the cover on it.

Those rockers look to be a good option. Some more work to make the holes, but if looks like the final edges then get neatly hidden by the outer flanges of the switches?

I totally agree on the VM. You are the only other guy I know of who has run this with a VM.

 

[yogi.1026]

Hi JohnH, I forgot to add that the hiss reduction in the HDR is worth the admission price alone, even my 'special' 7025 ax7 tube didn't completely get rid of this. Very quiet with the attenuator. I also may have been a bit harsh on the DSL, having a chance to play with the settings brings out the best in this amp, definitely worthwhile. On LDR I found using the 14Db slightly better as this channel needs to be cranked on 21Db. I dont know if they all do this or it's just mine?(big vol difference in modes)

 

[matttornado]

Hello again.

With no load plugged into the speaker outputs, I measure 30 ohms across the input. Is that correct & Ok if used as a load box only?

Looking at the diagram I posted of the Line Out, it appears the the ground goes straight from the attenuator's input straight to the Line Out output (purple wire).
So if I use a plastic output jack like all of the others, the ground should not touch the enclosure. The pot is not connected to ground at all from what I understand by looking at it?

 

[JohnH]

Hi @matttornado

30 Ohms would be the expected reading with no speaker, and all attenuation stages off. But to use it as a load box it needs to have all stages on, to max attenuation. Then you should find that the input reading is close to the nominal 16 Ohms even with no speaker.

The pot case is not connected in the circuit, so that can mount to the chassis with no issues

 

[matttornado]

I wired up the line out like the one I posted but it did not work correctly. I finally figured it out. Lug 3 must not be connected (grounded) to the pot & the ground from the attenuator's circuit gets connected to lug #3 instead.

At first I connected it right from the amp's input but it was very powerful so I moved it to right after the reactive stage instead and that tamed it a lot.

Also, I believe an audio taper pot would work much better, making a more of a gradual increase in level, & would work fine connected to the amp input jack.

 

[JohnH]

Thanks for catching that, and I agree about the line out. Sorry I didn't spot it myself!

 

[tmingle]

Thanks tmingle! Is your attenuator circuit like the one mentioned in this thread w/ isolated ground with plastic jacks?
Did you just use a metal jack for the line out?

Yes. Metal jacks for all.

 

[tmingle]

Hi @matttornado , that's great news and an awesome looking build! The resistance readings are totally looking right on the money given normal tolerances. I'm really glad that its working so well and I appreciate how carefully you have built it and I thank you for trusting the design.

On the temperatures, you are cranking a 100W Plexi and I reckon that attenuator is just cruising nicely. If you can touch the resistors for a second, then it would mean they are considerably less than 100C, eg, as an example of 100C, you cant touch a metal pan of boiling water without pain?

When it gets a lid, vents are needed and air in through the base and out through the top is best. but I also think that self-powered fan idea could be a very good one to explore too, for a 100W build. Need to experiment to set it up right. I think it could be set so it only starts turning if the power gets up above say 30-50 W.

The line-out circuit looks right, and you might like to try it either at the input or the output. At the output, obviously its then on an attenuated signal but it'll pick up some more of the characteristics of your speaker, particularly the low bass resonance. But, we also know from @tmingle 's build that it works at the input too. At the input, you then have the option of not using a speaker at all, just set it all to max attenuation and use it as a load box. At -31.5db, the amp won't see any difference.

I use the line out into impulse responses & have tried to match the miced sound. It can get pretty close with right impulse. I do get better results with some IR's that I purchased vs some of the free ones that are out there. I am trying to learn sends etc. in Reaper to allow for adding reverb etc. I do believe that with the right setup that you can actually create a virtual room & place the speaker wherever you want it in the room. I suck at it, but it beats watching TV.

 

[6StringStewie]

Id like to ask for a bit of help from anyone building the attenuator shown above, either now or in the future.

...Any interest?

Hi John. I am interested in building an attenuator for my JCM2000 TSL122 (100W 2x12 combo.) The amp has two Celestion V-Type speakers (8 Ohms, 70W each) in series. I would like to attenuate to 20W, 1W, and possibly a 1/2W output. I read that you have a spice model that you can run numbers in. Do you need any other measurements to run spice for my amp?

Would there be any modifications to values needed if I replaced the 70W speakers with 50W-60W speakers?

Looking forward to a fun project!

 

[JohnH]

Hi @6StringStewie
I've got models all set up. But what I reckon would suit you would be a 16Ohm version of design M (tested many times), or the recent M2 (simpler, should perform the same but not tested yet!). If you don't need it, you can miss the full bypass switch. So from 100W, you'd get 20, 9, 4, 1.8, 0.8, 0.35, 0.16, 0.07 W using three switches. Might as well have a nice consistent range of equal db steps.

Or if you really just want fewer and specific steps, then two switches and I can work out values.

The speaker choice doesn't really affect the design, so long as it adds up to 16Ohms . But power ratings should be x2 compared to the 50W versions. I'd think though, that with a TSL, it's not needed to crank it fully to get it's best tone.

 

[6StringStewie]

I finally got through the whole thread. I will have a notepad and pencil ready for the 2nd reading.

I am curious about Attenuator M3 with the bass resonant circuit. As I understand what you have posted, the bass resonant circuit is feeding the amp the bass response it is expecting to see of a speaker directly attached. Is that correct? What is the difference in behavior of an amp with and without the bass resonant feedback? Will this correct the bass dB loss at the speaker? Would it be cheaper to build a bass boost circuit instead?

I found this Jantzen toroidal inductor that looks like it might work.

Stewart

 

[6StringStewie]

I have a question about the M model and the heat generated by R1. I could either get a 200W or greater 30 Ohm resistor, or use two 100W or greater 60 Ohm resistors. Let's say I get two 250W 60 Ohm resistors with heat sinks. By dividing the circuit between two resistors of a larger wattage, would this help tame the heat and keep the enclosure cooler?

Stewart

 

[JohnH]

I finally got through the whole thread. I will have a notepad and pencil ready for the 2nd reading.

I am curious about Attenuator M3 with the bass resonant circuit. As I understand what you have posted, the bass resonant circuit is feeding the amp the bass response it is expecting to see of a speaker directly attached. Is that correct? What is the difference in behavior of an amp with and without the bass resonant feedback? Will this correct the bass dB loss at the speaker? Would it be cheaper to build a bass boost circuit instead?

I found this Jantzen toroidal inductor that looks like it might work.

Stewart

Whether that bass resonance circuit is worth doing is still moot. Here are my thoughts so far about it:

The basic circuit (ie no bass res circuit) already corrects the output at the speaker to give a close match to a full-volume bass response, for all signal levels that are clean or not very distorted. This happens by the natural interaction of your real speaker with the resistors. So there's no benefit in adding EQ, since it's already basically right. You can see the bass peak in all output levels at the speaker.

But if you hit a bass note hard at high volume, it could saturate the output stage adding extra bass harmonics and compression, which then will decay dynamically. This is enhanced by the high bass resonance impedance, which will draw more from the amp at these frequencies. This may be audible, or it may not be. Design M3 should capture this accurately and feed it all through to the speaker. Design M or M2 will feed through a cleaner bass note, still of appropriate volume but with less bass flub/ harmonics. My belief is that M3 is more accurate, but M and M2, which will draw less from the amp at bass frequencies, will be like having a bigger more solid amp output with more headroom, for the bass frequencies. Ie, it could be a good thing.

The bass circuit is only active up to say 150Hz, and mainly only around a very few notes on the low frets of low strings. It's irrelevent if not playing those notes, and I suspect that, given the basic response is right without it, those bass harmonic variations are to some extent a quirk, and not fundamental to the character of the guitar amp tone which is all about low mids, mids and highs

As you've seen, the inductor and cap are not cheap. You may be in for $100 to $150 extra just to get them, whereas with careful shopping, everything else can be obtained for less than $100, soup to nuts!

I've not tried the bass resonance circuit myself (yet, but I might!). What I want to try first is to do some recorded tests on the low bass response first, to try to hear if there is any worthwhile difference between full volume and attenuated, using my Design M build. I'll need to figure out the bass resonance of my speakers, and then play some loud notes that try to provoke it.

But what we do know is that if you don't bother with the bass circuit, you can still build a great and very simple unit that you will like!

 

[JohnH]

I have a question about the M model and the heat generated by R1. I could either get a 200W or greater 30 Ohm resistor, or use two 100W or greater 60 Ohm resistors. Let's say I get two 250W 60 Ohm resistors with heat sinks. By dividing the circuit between two resistors of a larger wattage, would this help tame the heat and keep the enclosure cooler?

Stewart

I think using 2x100W resistors to give 200W capacity is a good idea. You can use two 15ohm in series or two 60Ohm in parallel, to get 30 Ohm.

In the wiring, you could place them not next to each other, using the space between for some of the less heated one's.

But I suspect that you won't need to crank your TSL to get it's best tone, since unlike the NMV Plexi's and also VM's, it's more based on preamp drive. The attenuator will help you get it out of first gear and cruising nicely

 

[6StringStewie]

I think using 2x100W resistors to give 200W capacity is a good idea. You can use two 15ohm in series or two 60Ohm in parallel, to get 30 Ohm.

I am thinking I might want to try two resistors in series. In theory, doesn't the 15 Ohm resistor run cooler because there is less impedance per resistor?

But I suspect that you won't need to crank your TSL to get it's best tone

Probably not, but I can't wait to try my amp on 11, lol.

 

[JohnH]

Either way that uses two equal resistors to make a combined 30 0hms will be fine, and share the same power equally. Power is volts x amps. Series connection halves the voltage on each resistor, parallel wiring halves the current, for the same net power.

 

[6StringStewie]

16 Ohm versions

... for M3, the capacitor value halved. This might make the cost a bit more feasible since one 100uF could be used...

Just curious to see if this would work and considerably reduce the cost of a build. JCM900 amps use 50uF+50uF dual can capacitors. Dual capacitors are two capacitors in one housing. Using both sides of the capacitor in parallel would yield 100uF (50uF + 50uF). These capacitors can be purchased for under $15.

Thoughts?

 

[JohnH]

Not sure but Id expect probably not. The cap needs to be bipolar with high current rating and low losses, otherwise it could overheat, and also not achieve the resonant peak needed. The kind of cap needed would normally be used in the crossover network of high-powered speakers.