If it works, how is it "too primitive"?
And for whom, the circuit or its owner?
In the end, all that counts is the effect. Do it however you want to do it, so long as it does the job.
Relatively speaking, I could now claim that most regulators are in fact primitive because the problem of line junk should have been dealt with long before it got to the power transformer. Better to prevent rather than cure (Serbian - боље спречити него лечити). All the regulator should ideally have to do is to keep the supply voltage constant throughout the operating range.
Last edited:
Too primitive means PSRR of 60db instead of say 90 db and 2ohms output impedance instead of 10 miliohms.
You are co-author of the very elaborate feedback shunt regulator , but suddenly you start supporting the simple possible regulators.Very well if you get personal satisfaction with them
Mains filters are not the only way to reduce line junk. UI and EI transformers reduce common mode noise effectively. Voltage regulators not only regulate voltage to desired value, but they also filter line and rectifier noise. Low ESR capacitors across feedback resistor makes very effective low pass filter up to high (line junk freq.)
I am the co-author of no such regulator. The actual authors were Zen Mode from the Serbian DIY forum and a friend of mine, Oliver Isailović. I did try it out and I did say I thought it was a very good reguator - complex, relatively expensive to make, but in my view, well worth the money, time and trouble. I have no problems with recommending it on basis of my own experience, I still have one blank board somewhere around here and may well use it again.
However, I am a co-author of another, far less elaborate shunt reguator, with my friend Oliver Isailović. That one I do use in my headphone amps/preamps, and I know I will use it again, because it does provide good results for the time and money invested in it. That regulator is limited to low power circuitry and will, at this point, handle up to about 350 mA of current with voltages of up to +/- 56V or so.
Therefore, my preference for "virtual battery" reguators (as advertised by Panasonic/Technics) is a long lasting love affair, now well over 25 years old. If done properly, it can provide excellent results overall. In my terminology, "properly" means Fisher & Tausche 2,200 uF caps (two before and 1 after the regulator) and comprehensive additional lower value caps.
I also have some classic text-book regulators, using discrete bipolar devices, very scalable and in some versions capable of handling even the current stages of a power amp, but my experience is that this can be almost duplicated by simpler capacitor schemes, however backed up with more massive than usual power transformers (e.g. if calc says 300 VA is enough, use 500 VA or bigger).
In my experience, once you split up the power lines of the input stage, the VAS stage and the predriver from those of the driver and actual output devices, your job is say 90% done. Therefater, you are well in the sector of diminishing returns, you progressively have to pile money in for ever smaller benefits. And often enough, you can't hear the difference.
As for the PSRR ratio, well, that depends heavily on your baseline input signal. If you have cleaned up your act beforehand, that 60 dB may in fact yield overall better results than another at 90 dB. Also, the PSRR is influenced by one's CCS circuit(s), which themselves (regarding the most usual types) vary from around 24 to around 86 dB - there was some excellent input in this topic some months ago, with attached schematics of the types tried. I do tend to use (relatively) elaborate CCS circuits, mixed BJTs and FETs in cascode connection, which should yield some of the best PSRR around, on top of PSUs already cleaned up before the signal ever even got to the power transformer, so I'm not worried.
Last edited:
BTW gold plated PCB is a real gimmick. Let us watch IPC, NASA etc. serious soldering tutorials, and we will see that gold plating must be removed from soldering surfaces by applying/removing solder by wick, gun etc.
Imagine soldering components in this way on Spectral, Threshold , Coda etc equipments. Silver in solder alloy is useless for gold plated surfaces.
Imagine soldering components in this way on Spectral, Threshold , Coda etc equipments. Silver in solder alloy is useless for gold plated surfaces.
As you say it, you are in fact using transformers as filters. Have you looked at the output impedance of a large EI transformer at say 100 or 200 kHz, in comparison with an equivalent toroidal transformer?
As for voltage regulators, well, I also have to use them anyway, so to me, their filtering function is a bonus.
Regarding low ESR caps across feedback resistors, I leave that to you, I like my feedback paths to be as simple (empty) as possible. I mistrust caps completely, they are the weakest ink in any system, I feel, and in terms of time, they are the first to start slipping. So I try to rely on them as little as possible.
Yeah, while sales were wild up to now ...
The local economic situation is so bad that for the price of the cheapest Spectral device a man would have to forego 100% of like 8 or 9 monthly wages, at least the lucky ones who do have employment.
The boards in question were fabricated by a shop in Santa Clara CA that was supplying Hewlett Packard with the same technology from the same specifications. HP did not abandon the gold plated PCB's until the '90s. They actually have better solderability than tin/nickle, especially after a few weeks. Most of the hi-rel parts used in military applications have gold flash on the leads to protect them. The shop that assembled the boards was run by the guy who wrote the NASA specs for assembly and reliability. I learned a lot from him. It was his retirement occupation. . .
(BTW, just imagine the cost of gold in a tank big enough to plate a 12" X 18" PCB. No wonder the technology was abandoned, too risky.)
Practice has changed since then, as surface mount became the dominant technology and it has very different requirements. Lead-free has really changed things (still not required in life critical applications for some?? reason).
The number of working to spec 30 year old products (HP and Spectral) made this way suggests its not a bad way to make things.
A wise, skilled man would build something as good or better for ten times less money.Yeah, while sales were wild up to now ...
The local economic situation is so bad that for the price of the cheapest Spectral device a man would have to forego 100% of like 8 or 9 monthly wages, at least the lucky ones who do have employment.
sorry to hear that - if you think it was teaching you something you now have a lot to of mislearning to undo
Geddes and Lee actually tested more than a couple of guys on a couch, employed actual psychoacoustic and statistical controls so you might start with the GedLee Metric papers
you might also really do the IMD math for complex music - you don't always have just single tones in music so the "consonant harmonic structure" gets spoiled by anharmonic sum and difference tones - even with pure 2nd order distortion
Last edited:
Sounds like an interesting challenge. Are you guys up for the task? If fancy metalwork is not required I suspect something almost as good (no 1 MHz power bandwidth requirement) could be built for under $30. Perhaps more if it has switching etc.
bimo, the chap behind the work is a recent member here, only 2 posts though! - http://www.diyaudio.com/forums/tubes-valves/240043-what-tube-sound.html#post3578810
- Status
- Not open for further replies.