It's all buried in that other thread. I don't have the measurements saved anymore. My set-up can measure down to roughly -120 to -130db depending on the the day and phase of the moon.
The parameters of normal measurement are simple though. I hook up the QA401 to the output of the amplifier with the ground clip near the signal ground and connect some kind of a load. Then a Play a sinewave and keep increasing the voltage.
Oh hey Tubelab.
The "clipping" I applied to the opamp was not true clipping since it was able to be corrected.
I used an LT1364 and put it under a 20 ohm load if I recall.
Then I drove the voltage up and got it as close to 0db distortion as I could get.
So it was more of a current clipping, I don't know the exact internal mechanism but I drove it at 3 times rated current before the distortion was unable to be corrected. At all other points however it was underneath the noise floor.
As far as that particular circuit goes I just want verification on whether the performance is as good as it seems so I can look for investors so I can produce and sell some stuff. That way I can continue my R&D without spending half my income and starving myself.
I got a backlog of cool designs I think people will really like but I lack the funds to build and sell.
That circuit seems to be a good catalyst to get things moving in that regard. Plus I've put so much time and too much of my money into developing this stuff, I have to at least recoup what I've spent, or at least that's the goal.
Sorry no, I meant it was not visible on the QA401 analyzer. I only see the noise floor regardless of load and power output. It's even stable under a fraction of an ohm, I was direct driving some ribbons with it earlier, it was great, I was not expecting that.
I have Victor's oscillar boards which are supposedly capable of -150db but in any case you're right. I need something I don't have, in one way or another.
Could you elaborate here? I don't quite understand what it is you are saying I should do.
Would love to hear your opinion on my tube circuits too, I know that's kind of your shtick.
Last edited:
Until you stated this, I had no idea what a QA401 was.
I did not know what the QA401 was, but I looked over the web site and the manual, and it appears to be just what I was saying. The QA401 IS a high end external sound box that has been designed specifically for audio measurements. Apparently there have been some upgrades since the manual online was published, since the manual refers to 24 bit capability, but the DAC / ADC chips listed in the spec sheet are 32 bit parts.
I am using a commercial USB sound module intended for music recording and playback with 24 bit 192 KHz capability, and running some software to make measurements.
I didn't see it spelled out on the manual (didn't read it all) but you should be able to perform a "loopback test." This is where you connect the outputs directly to the inputs and make distortion, noise, and frequency response tests with no amplifier present, and any attenuation set to minimum.
You should get THD readings below -100 dB a frequency response flat from a few Hz to nearly 96 KHz, and a noise floor well below -100 db.
Then hook up your amp and make measurements. If you still get readings below the measurement floor, you might have something, but making real measurements in the below -120 region will take some effort.
There is a thread on here somewhere concerning measurements and an amplifier with numbers in the below -100 range, but I am of the assumption that extra zeroes in the distortion percentage don't really matter, so I haven't really paid much attention.
if you are going to share, do it a 100% or none at all, otherwise your thread would properly belong to the vendor's since you have a commercial stake...
this is a diy forum where members share ideas unconditionally, except perhaps where a group buy of boards happens...
this is a long standing tradition of the this board who had stood for almost 20 years now..
this is a diy forum where members share ideas unconditionally, except perhaps where a group buy of boards happens...
this is a long standing tradition of the this board who had stood for almost 20 years now..
Oh yeah, I did all that already. Same results.
Well, crap. I guess I really have no choice but to shelve the project then.
I say this without malice; I think you need to do a lot more work on your testing and documentation if your goal is to convince anyone your technology is worth a further look, let alone significant investment.
Nobody wants to dredge through pages and pages of waffle to get to what should be pretty standard test results for an audio amplifier.
Nobody wants to dredge through pages and pages of waffle to get to what should be pretty standard test results for an audio amplifier.
Doesn't matter what I do people will question it until it's measured by a reputable third party and I'll continue to waste my time. But that will likely never happen.
It's a never ending tread mill that I can no longer afford. I'm out. I tried. I'm done. Thanks, to those who weren't a**holes to me.
It's a never ending tread mill that I can no longer afford. I'm out. I tried. I'm done. Thanks, to those who weren't a**holes to me.
Plenty of labs out there who will do the testing for you and are independent.
Perhaps worth borrowing some money to pay for this, if your designs have commercial merit. In the end, you have to back yourself before others will back you [emoji106]
Perhaps worth borrowing some money to pay for this, if your designs have commercial merit. In the end, you have to back yourself before others will back you [emoji106]
That's exactly my own method, except I use what was supposed to be a "hi end" broadcast sound card (PCMCIA in fact) which has the huge advantage as a card bus of being hotpluggable direct into the PCI bus.
When I run loop back tests, the sad fact is what is claimed is very rarely true in the real world, and that applies to loads of them.
We have this all the time in the studio, with differing standards Dante, Ravenna, IP networks, and the insistent compatibility problems, lots of bugs, & DACs that don't do what they say.
In the end Merging tech came along, but it's a 10 000USD+ solution, and has an "elephant in the room" sort of bug, which I have to fix fast. 😱
There is a forum on here for a good distortion analyser, all you have to do is LOOK!
How many times have you seen a computer motherboard claiming 100+ db dynamic range from the on board audio? I have one of those boards and with a 6 inch jumper plugged into the jacks I see about 70 db since the noise is so high.
For years I used a M-Audio Audiophile 24/192 PCI (not PCIe) card that measured in the mid 60's and was flat from 10 Hz to around 90 KHz. Unfortunately most of my PCI based computers have died.
I recently got a Focusrite Clarett 4 Pre USB sound box. I use it for recording, and have not performed a loopback test yet.
In the world of High End Audio, ANYTHING that is DIFFERENT from the accepted norm will be scrutinized and criticized heavily, even with valid measurements.
I was a development engineer at Motorola and trained to use the best component for the job, whatever that may be in a product where people's lives depend on it working properly (mission critical public safety radio equipment).
I built tons of tube stuff, HiFi and guitar amps in my youth, but became a mostly solid state guy when I got hired at Motorola in 1973....mostly because silicon was FREE just by filling out a sample request form.
In the mid 90's I started making tube stuff again. I set out to build the ultimate SET amp in the late 90's and after a few years of tinkering I had a design that worked.....BUT, I had committed the ultimate tube audio sin....there was a mosfet follower and a CCS chip in the signal path.
I made myself one, and a friend heard it, and wanted one....This repeated often enough that I couldn't make enough of them. Most of my friends were coworkers at Motorola and of a technical mindset, so I had some PC boards made, printed up some simple instructions, and let people build their own.
This continued until one of my buddies said, "why don't you put this stuff on the web?" This is how Tubelab got started, but almost died. The initial reaction to the schematic was overwhelmingly NEGATIVE. People posted stuff like, "You should change your name to Transistorlab." It took a while but as the first few builders started talking about how it SOUNDED, things changed.
Now, 15 years later mosfet followers and CCS chips have become somewhat mainstream in tube designs. That doesn't mean that they are universally accepted though. Other hybrid designs are still mostly frowned upon.
Tubelab Inc started as a hobby, but remained an independent entity since incorporation in 2005. It has been able to pay for itself, but has never made a bunch of money. I made it a point to never borrow money for Tubelab, but there were some lean years where it did go slightly negative. My career at Motorola ended when the plant closed in 2014. When Tubelab fails to pay it's own bills it will go away.
The high end audio market is shrinking. Most of today's generation is happy with an iPOD playing MP3's. The vacuum tube corner of that market is shrinking even faster. Would Tubelab, or any other NEW start up be successful today? I doubt a new tube HiFi startup would have much of a chance unless they leveraged some sort of market advantage.
If you are really serious, and do believe that your designs have something to offer, then get some prototypes built, and into some vocal users hands. I was fortunate enough to work in a plant with 5,000 other somewhat technical people, and critical mass could be achieved by word of mouth, often by loaning out my amps.
In my youth I got noticed by loaning one of my DIY guitar amps to someone at a local college, the University of Miami, where their music program was large. It wound up at one of the frat houses, and soon I had request's for more.
Tubelab got a boost when someone took their implementation of one of my SSE designs to the Burning Amp festival.
You have a high end audio analyzer. Get some plots of the thing in loopback mode, then show that your amp can do just as good, but at a much higher power level, make a video, then put it on YouTube. Have some sort of marketing plan in place before doing this, in case it works.
I have received several emails just because I was bored one day and shot this crude video of a prototype with my phone. That video and a few forum posts have sold maybe 30 boards.
YouTube
There are ways to get your product out there, but it is up to you to figure out what you want to market, and how to present it in a neutral way so that people come to you for it.
In my case, I post my schematics and design theory, then sell a blank PC board for DIYers to make their own. I did offer parts kits at one time, and may do that again, but it involves some constant scrutiny and attention that I couldn't provide at the time when our parents were dying and my 41 year engineering career was ending. Selling completed amps involves a lot more money, liability, and risk.
Sometimes there's so much negativity and discouragement, it's refreshing to see these types of posts from someone who's had their share of what life puts in front of them, and yet that person is here encouraging others onward. Thank you.
Since you are using LTSPice its possible to measure THD of your design in simulation mode.
I did it myself some years ago and yes, numbers are very very close to reality.
IMHO this is the most useful technique before going into the prototyping stage.
LTSPICE THD Analyzer
Audio Total Harmonic Distortion Analyzer for LTSPICE, making THD vs. Amplitude and Frequency sweeps in LTSPICE | Audio Perfection
Spice results for me have been pretty accurate to what I've seen on the bench as well. The EC method I discussed on the last page showed the same results in spice too, among other tests. I believe I also posted the comparison between spice and reality in that test on the other thread.
I've done all manner of spice distortion tests. They all say the distortion is below -300db. It won't even register it.
The one here is one that I took at the mid point between the feedback resistors of an inverting amplifier.
The results are always the same. Whether in sim or on the bench.
The most typical and convenient way I like to measure spice distortion is by doing a transient analysis and looking at the voltage distortion between two feedback resistors in an inverting amp. I usually end up with low nano to mid pico volts in the design discussed, depending on what I attach to it.
Yes, Tubelab is a breath of fresh air. I've always liked him for his down to earth nature.
Whenever I discuss anything here I feel like I'm drowning and my efforts are wasted.
Case in point, no one has even mentioned the circuits I made the thread for, even though they are a step up from the typical circuits of their kind. Clearly no one is interested. Dunno why I bother.
Granted I made those circuits a long time ago but I put a ton of effort into them at the time.
I've spent over 20 grand in my R&D endeavors getting to this point and thousands of hours while barely scraping the edge of homelessness with no help, can't even afford a car. The frustration is continuous.
In any case I've done the loop back test Tubelab is suggesting tons of times..I'm pretty sure I've posted the results of them somewhere in the other thread.
I don't have the money to do a high power test, the highest I've done is about 60W and I accidentally fried the output stage on that one so it's no longer usable.
I don't know how many watts I need to do to "satisfy" people.
Starting at the beginning of next month I no longer have the money to spend on any of this stuff for the foreseeable future. Unfortunately I am completely obligated this time, so what I have is what I have. and I've now been soft banned from discussing it anymore on here anyway so yeah, I'm done.
Last edited:
I had a struggle to find enough thumbs up to put across my real feelings about this post which says it how it really is.
so there you are I have watched and admired lots of things you presented over a few years.
I used LYNX sound cards, which are better than *,
To us sound guys turned out good, (I think plain luck) one of which even works on my now ancient DEC Alpha 64, so I even played the OS cards in and out of Linux to try to cure the worst aggravations.
You still have a job to convince a "hi end" audio shop that a professional PC sound card DAC is really better than their awful vinyl experience.
If you think "hi end" audio is bad, just look at what happened with the IT industry and Windoze, then the sheer lunacy going on with speaker cables, and the other conmen selling power conditioners and power cords...with "tube rolling", Asian extremism, and NOS crazy price speculation thrown in.
I starts to think making a go of anything these days is a combo of stubborness, sheer luck and very little technical virtuosity.
Risk is everywhere in that stuff, some of it quite high.
I really wish it were not so.
I would be sorry to see Tubelab disappear it's one of my 5-6 references in the world you can count on one hand, of people who actually know what they are on about.
I took on an RF contract engineering job in late 2018 and early 2019 that put enough funds into the Tubelab bank account to get through 2020. I will be 68 years old then, and I can't do this forever, but there should be a few more years left in me if the financials look OK at the end of 2020.
Without that cash infusion I know that 2019 did not generate enough sales revenue to pay all the bills. The extra cash will allow for the launch of one new product in 2020. I have two in the prototype stage. The results of some testing that I am setting up now may help decide which of the two gets the nod....I may also take a poll on the Tubelab forum.
UH, I have spent most of my time since November upgrading all the networked W7 machines in my house and most of Sherri's family to W10. There's a few more to do as long as the undocumented free upgrade still works.
Motorola had a good staff of in house IT guys until the bean counters decided to outsource it all. What a joke that was. The Dell on my desk died, I diagnosed it to a dead memory DIMM. I told the IT guy who took my PC that. After two days they bring it back with one of the DIMMs removed (exactly what I told him to do), and tells me that a new one is on order from Dell. About a week later a different guy shows up in my office with a DIMM in hand who admits that he has never seen one before, and has no clue what to do with it. I gave him a lesson in taking a PC apart and plugging in a DIMM, He had clearly never seen the inside of a PC before. These guys were running the network in a company that was constantly under cyber attack for the technology it created.
When MS released a bad update to W7 it killed about half the PC's in the plant. Total chaos ran rampant for over a week. I laughed since my old Dell still ran XP.
There is a good deal of serious con games in any industry where some serious money is out there to be made. I can think of several, but the diet / fitness / nutritional supplement industry has to be the worst. Don't forget the free energy weirdos that won't go away.
When you are working with tube equipment where there are often no zeros before the significant figures in a distortion percentage, any sound card even a 16/44 is good enough to make valid measurements. I got started in the 90's with a Media Vision Pro Audio Spectrum 16 sound card. With some careful wiring and grounding to avoid PC noise, you could get a dynamic range well into the 70 db range. Granted some of the early filters created some weird aliasing when overdriven and sometimes made dynamic percussion sounds appear "metallic," but we have come a long way since then.
Let me give you a hint at least on this thread before it gets polluted by the bizarre meanderings of our Ozzie friends...(on the other)..
All this kind of work revolves around compromise and pragmatism.
It's the same in another largely unrelated engineering field I have done a lot of work in.
Let's say it's "horses for courses".
At the end of the day what gets the bills paid, is not how small a distortion is available at the amplifier end, or how good or bad the source audio feed is (about 90% of all commercial recordings are absolute toilet...),
or how huge a distortion is caused by the vast majority of speaker drivers, and how Kellogg's invention has scarcely evolved in 100yrs, and how the vast majority of speakers are not able to resolve even 16 bit audio particularly at the LF end....
I did mention the few people I consider who actually know what they are on about...and recommended you to speak to Jack E at Electraprint, or Gillespie who lives on another forum, or Jantszen who's father INVENTED the electrostatic loudspeaker...
They are all very busy people, but with vast experience who will put the grey matter into place between your 2 ears, telling you what SOUNDS good, and what MEASURES WELL, are 2 totally different things.
That btw is why 100s of chips and opamps are used in the chain that gets music to you, some of them from ancient Neumann valve mics, some via ancient valve based equalisers, or who knows what else, to emerge usually from a DAC.
Ie. at the end of the day SPICE is rather irrelevant but IMD is not.
Spice like all simulations, in fact same as the weather on the TV, is not the same as what you see out of your window.
Science has to some extent helped, but,-
it's not what you know that causes problems, it's what you don't know, and the infamous "unknowns you don't know".
If you insist on ploughing your own scientific furrow, having spoken to people with decades of experience, the risk is entirely yours.
I doubt I could be fairer than that.
tubelab is one "shirtless dude" who knew his stuff and shared a lot more than he should...
like the rest of the membership with solid "ethics" he got a paid for forum so he can strut his commercial stuff, he knew what diyaudio was all about and knows how to get along well...he will never take advantage and get something for nothing....
there are others like him, who are on their way offering value products...
like the rest of the membership with solid "ethics" he got a paid for forum so he can strut his commercial stuff, he knew what diyaudio was all about and knows how to get along well...he will never take advantage and get something for nothing....
there are others like him, who are on their way offering value products...
I think there is a problem with the idea of a -150 dB amp, and 140 dB EC, giving a -290 db result.
Extreme Op Amps do get tested by killing the gain by so many dB at the input differential stage and measuring the results, then summing the two. That is OK as long as the results do not drop into the noise floor from the feedback resistors. Feedback theory assumes the passive resistors can be perfectly linear, so loop gain is the controlling factor.
However, EC is an active device feedback loop. It is not infinitely linear like ideal resistors. In this case it is putting artifacts into the feedback signal at the -140 dB level. When you combine this with a -150 dB Amp, then -150 dB artifacts and -140 dB artifacts are being combined and the EC will only be able to fix the result to -140 dB still.
Worth checking anyway before you spend your last $$$ on it.
There is another way to test distortion than the usual method. I have been playing with it lately for a tube distortion visualizer (mainly to see crossover for class AB). I'm not sure it can do any better though, but it might be possible to optimize it for a special measurement, like the Op Amp tests do.
What I do is use a very linear ramp or triangle signal for signal input to the DUT. Then the DUT (Amp) output is differentiated. The derivative of a ramp signal is a constant. Any variation in gain then varies from the constant. An X-Y scope displays the "constant" derivative output versus the input signal. Giving a panoramic display of Amp gain versus input signal voltage across the tested range. Perfect for crossover distortion analysis.
To increase sensitivity, one could use a very steep, but limited range, ramp at each desired input voltage test point. This will make the derivative into a larger signal. Then test repeatedly at all relevant signal V test points. Trying to get past -150 dB may prove to be daunting still. This could all be done with some special software using a 24 or 32 bit soundcard to generate the ramp bits (and an attenuator before the Amp input to make the bits "smaller") and read the derivative bits.
Extreme Op Amps do get tested by killing the gain by so many dB at the input differential stage and measuring the results, then summing the two. That is OK as long as the results do not drop into the noise floor from the feedback resistors. Feedback theory assumes the passive resistors can be perfectly linear, so loop gain is the controlling factor.
However, EC is an active device feedback loop. It is not infinitely linear like ideal resistors. In this case it is putting artifacts into the feedback signal at the -140 dB level. When you combine this with a -150 dB Amp, then -150 dB artifacts and -140 dB artifacts are being combined and the EC will only be able to fix the result to -140 dB still.
Worth checking anyway before you spend your last $$$ on it.
There is another way to test distortion than the usual method. I have been playing with it lately for a tube distortion visualizer (mainly to see crossover for class AB). I'm not sure it can do any better though, but it might be possible to optimize it for a special measurement, like the Op Amp tests do.
What I do is use a very linear ramp or triangle signal for signal input to the DUT. Then the DUT (Amp) output is differentiated. The derivative of a ramp signal is a constant. Any variation in gain then varies from the constant. An X-Y scope displays the "constant" derivative output versus the input signal. Giving a panoramic display of Amp gain versus input signal voltage across the tested range. Perfect for crossover distortion analysis.
To increase sensitivity, one could use a very steep, but limited range, ramp at each desired input voltage test point. This will make the derivative into a larger signal. Then test repeatedly at all relevant signal V test points. Trying to get past -150 dB may prove to be daunting still. This could all be done with some special software using a 24 or 32 bit soundcard to generate the ramp bits (and an attenuator before the Amp input to make the bits "smaller") and read the derivative bits.
Last edited:
I don't disagree with anything you just said. My personal opinions on what sounds good are not representative of my interest in getting that particular circuit verified. It's just that the results are too good to ignore and the price/performance ratio of the circuit is off the walls. Not something to ignore to someone that is hopeful in starting an audio business.
Agreed. I should have more carefully documented my experiments. But it's clear to me that no amount of logic or evidence is enough to convince anyone as long as it comes from my mouth.
I need a reputable third party.
It's not really an EC, I just used it as one. The fundamental way it works is by adding an absolutely bonkers amount of gain into the main feedback loop without becoming unstable. The stability of it is really the main specialty here.
The 1khz voltage distortion in spice is unregisterable in transient analysis and AC analysis. I have to use 20khz to even see it.
So essentially what I did during the opamp test was add the circuit in question into the feedback loop of the opamp and used the opamp as the output device. I can adjust the amount of gain added so it is effectively "correcting" the errors of the output device. You can let me know if that changes your suspicion, I am not an expert on feedback theory.
Unfortunately this is beyond me.
Last edited: