This thread is over 100 pages now and the search is not good for that. If you are talking about turning the Mod-86 into a myref style current amplifier I suspect you would have to do all that yourself.
I'm sorry to hear that my science-based approach comes across as a disagreement. That's unfortunate...
I don't recall any mention of the input impedance. The input impedance of the Modulus-86 and Parallel-86 is 48 kΩ differential; >3.2 MΩ common-mode. I really don't see a problem here.
I also don't recall any mention of your criteria. You've asked lots of questions about capacitor substitutions and I provided answers. In most cases the answer was, "I highly recommend that you populate the boards according to the BOM". In the vast majority of cases, the substitutions would have made no difference in sound quality at all. That's due to the extremely large loop gain of the composite amplifier topology. In some cases, the substitutions would have affected stability, which would have made the performance worse. I will not recommend those substitutions.
You also mentioned that you planned to replace the 0.25 W feedback resistor around the LM3886 with a 0.125 W type, regardless of me pointing out that the resistor would not be able to survive the dissipated power under high output power conditions.
Sorry, dude. I can't take that seriously. I deliver quality products that are well-engineered, well-designed, well-tested, and well-documented. I will not recommend that builders populate the boards with components that will not survive rail-to-rail operation of the amp. That would be poor engineering and downright reckless. That's not my business model.
I am always open to suggestions for improvement. However, please back up your claims with data. Show me how much improvement I can gain by following your recommendation. Show this based on a simulation, calculation, or at least an explanation rooted in circuit theory. Fact-based suggestions are always welcome.
Your preference seems to be to drive the speaker using a current source. The reasons for this are unclear to me, but whatever... If you want an ideal current source, the Modulus-86 is a rather poor choice as it is as close to an ideal voltage source as you can get. You may also want to read up on Norton and Thevenin equivalent circuits. It turns out that any combination of current sources and voltage sources with their respective output impedances can be reduced to one current source (Norton) or one voltage source (Thevevin) with one output impedance. You can also convert from a current source to a voltage source with some simple math.
Ohm's law still applies. If 1 A is driven through an 8 Ω impedance, 8 V develops across the impedance. Similarly, if 8 V is imposed on an 8 Ω impedance, 1 A flows through it.
Some will probably point out that a speaker generates back-EMF and that this should be dealt with. That's true. In my opinion, the back-EMF is an unwanted error signal that should be eliminated. The best way to do this is to shunt it to ground, which is why I use a power amplifier with a voltage output. This allows any back EMF to dissipate without disturbing the voltage to other drivers.
A voltage-drive amplifier with low output impedance will have very tight control over the speaker cone. If your speakers aren't up to the task, you'll probably find that the break-up behavior of the speaker cones becomes more pronounced with a good voltage-drive amplifier. This is because the amp has tight control of the speaker cone. This is a desirable trait, right? This is what we want, right?
This is one of the ways an amp exposes a bad speaker. Unfortunately, for some, the reaction is, "I like my speakers. They are flawless. It must be a bad amp". If these people would measure (concept!) their speakers, they would realize that the wiggles in the frequency response accompanied by a rise in THD is caused by cone break-up that gets excited more by a good amplifier.
This is why I take a fact-based, science/engineering-driven approach to circuit design. This approach allows me to isolate issue in the playback system so I can deal with each issue individually and achieve better sound quality.
Faith-based subjective tests do not provide this insight. They may be able to point out an issue in the playback chain, but relies on trial and error to identify the cause. In many cases, such as in the bad speaker vs good amp above, the subjective approach leads you down the wrong path.
~Tom
Measurements of input impedance of pin 2, pin 3, and between pin2 and pin3 over the audio spectrum.
Measured data the reflects damping capability over the audio spectrum.
Measured data or simulation of current and voltage gain and phase showing compensation for driver making the system phase and gain linear.
Would like to see the above.
You language reminds me of an old friend Mr. Dieckmann in TX.
Oh! There is a nail.....
Measured data the reflects damping capability over the audio spectrum.
Measured data or simulation of current and voltage gain and phase showing compensation for driver making the system phase and gain linear.
Would like to see the above.
You language reminds me of an old friend Mr. Dieckmann in TX.
Oh! There is a nail.....
Hi
I've been listening to mine for the last 4 hours... I have 29.7v going to the Modulus-86.
No pre-amp, playing through my Sachiko horns with FE206EN's (Usually powered by a 300b tubelab TSE). They sound great, really great!!
I'll soon test it through my other speakers (FE208EZ BVR's).
I'm using some expensive heat sinks in mine so the complete price would be around $600-700 (excluding the case). My heatsinks are $400-500 (0.25w/k meant for the Zen/F5 in a small case so they barely heat up with the fans disconnected), a more realistic price for a complete build is probably closer to around $300 + case... I think it's excellent value for money and you probably won't find any better...
Excellent design, well done thank you Tom.
The short answer is that there are none. The best speakers have 3 (!) orders of magnitude more distortion than the Modulus-86, mediocre ones 4 orders of magnitude more.
As to linear distortion, there are some loudspeakers that have fairly flat fr performance, but still nothing close to what amps can do.
DIY, there is no hope you can approach the quality level of the best JBLs, for example, for the simple reason that the drivers you need for that kind of performance are not available to hobbyists.
You can find spot measurements in the THAT1200 data sheet. You can deduce from the equivalent circuit schematic that the differential input impedance is likely flat across the audio range.
The common-mode impedance depends on the output impedance of the bootstrap driver. There's data in the data sheet for 60 Hz and 20 kHz. Based on how those circuits work, you can likely reconstruct the impedance curve by assuming a +20 dB/dec slope at 20 kHz and use the 60 Hz number as the "impedance floor".
The damping factor depends on the load impedance. DF = Zload/Zout, so what you're after is Zout. I did measure that on MOD86 R. 1.0. The damping factor of the MOD86 is about 50 (8 Ω) across most of the audio band.
I'll take the measurement for MOD86 R2.0 and PAR86 and post it here. I have a few other things in the pipe that I need to execute on before I can get to this measurement, however.
So you require the amplifier to be adaptive and automatically adjust its characteristics based on the speaker driver connected? Good luck with that.
~Tom
Amp features that DBLTs have shown to sound better.
http://www.diyaudio.com/forums/solid-state/230492-audio-power-amplifier-design-book-douglas-self-wants-your-opinions-228.html#post3846750Thanks Bill.
You are the first person on this forum who has expressed interest on stuff that DBLTs have shown to result in better sound.
Makes sense. Most impressive sound at home I ever had was a combination of Radford STA-25 and celestion Sl600 when I had a decent sized room (final year at Uni, with a friend lucked out on digs). I had the SLs on demo and had to return them as they were too good. The combination was not neutral by modern standards, but musically very satisfying. Of course have no idea what overload recovery was like on the Radford. Do need to restore that one day. Oddly the Radford didn't review well in America, but is much loved in UK and Japan. Go figure!
The classic valve amps from LEAK, Radford, Dynaco & QUAD were unconditionally stable into any load and had exemplary overload behaviour & recovery ... unlike many Golden Pinnae offerings, both ancient & modern.
Their designers are quite contemptuous of the Golden Pinnae brigade whose amps often (always?) have flaky behaviour into real speaker loads. I had the privilege of working with Ted Ashley who designed the last of the great LEAK amps and have had several robust discussions with Peter Walker of QUAD on amp & speaker design.
Though no longer involved with day to day design at the time, they were both keenly aware of the latest designs and theory and always had very clear and well thought through practical reasons for their individual approaches.
It's very illuminating talking to the great designers of old and learning where THEY thought they had done well and where they thought they might have done better. Also the modern bits they wished they had available in da old days .. its NEVER 'hand carved from Unobtainium by Virgins stuff'.
Look forward to whatever measurements you can provide. I have read a lot about the THAT chips, but have never seen the measurements that match the claims. I would prefer to see measurements with the shortest cables used. Probably less than 2 feet.
Based on what I have measured with the Hafler XL280 and the MyRef, I doubt that Zout is what I am looking for. But it make no sense to argue about it.
Last edited:
This is why I am not convinced that distortion isn't the holy grail for amplifier design, it is icing on the cake when you have other things going right. Which is also why I feel considering the audio system as a whole is a more appropriate approach for modern technology.
Why don't you tell us what you looking for .. in words that allow one to design for it.
I have compared damping measurements as described in the MyRef thread, would like to see how damping compares with the same type of measurement.
I have also expressed the what I think is important in terms of phase and current.
Constant input impedance measured is also a requirement as explained before regardless whether balanced or unbalanced operation.
Under these basics, I do not see any interest, so it makes no sense to say more.
I have also expressed the what I think is important in terms of phase and current.
Constant input impedance measured is also a requirement as explained before regardless whether balanced or unbalanced operation.
Under these basics, I do not see any interest, so it makes no sense to say more.
As I posted in Post #1048, the damping factor, DF, is a function of the output impedance.
DF = Zload/Zout, where DF is the damping factor, Zload is the load impedance, and Zout is the output impedance of the amp.
You say you don't want Zout, but expect DF to be provided. That request makes no sense.
~Tom
DF = Zload/Zout, where DF is the damping factor, Zload is the load impedance, and Zout is the output impedance of the amp.
You say you don't want Zout, but expect DF to be provided. That request makes no sense.
~Tom
It is because I have not seen the damping factor as per your definition to relate with how it will sound. From the measurement methods described in the MyRef thread, I replicated the measurements and clearly know how the curves will relate with listening experience. I used a real driver in those tests.
Last edited:
What damping factor are you looking for then? How am I supposed to provide measurements to your satisfaction when you redefine commonly used terms without providing your definition?
~Tom