I once ruined a perfectly nice amp with Schottkies. Didn't have enough heater supplies and real estate for the two 83s from the prototype and thought: how bad could it be? After a while my ears will get used to the Shottkies anyway. Sadly it didn't happen.
Sometimes i think i like tube sound just for the rectifiers. Should try feeding an opamp circuit some day.
You have to measure and listen, once you have what you feel is a viable design you validate that in the real (audiophile?) world. You set up listening sessions with unbiased, experienced listeners, either in their systems or yours. When possible you provide a known, respected competitor's unit for comparison purposes. You loan prototypes to trusted clients, friends, dealers. Even for a one off diy project you can do most of these things along the development path, but the important thing is the engineering content goes in first, changes then occur as a result of listening. You make an attempt to rationally correlate what people heard with what you can measure, and then you make improvements if necessary, or come to an understanding of what can be accomplished sonically within the original goals of the design. (Sometimes the goals aren't that reasonable or well thought out, and need a rethink.)
It helps a lot (IMHO) to listen to live music in an acoustically good venue (classical/jazz/acoustic - and even some rock if the PA system is not over cooked) so as to have some frame of reference as to what live music sounds like, and though no recording ever fully captures the essence of a live performance it does help to know what things are supposed to sound like. In some instances this may be a better frame of reference than comparing to someone else's interpretation of what sounds right. This is particularly true if you are trying in any way to recreate some ressemblance of the sound of a live event in your listening space - which incidentally is one of my primary goals for appropriate source material.
When I was in the commercial sphere my amplifiers were frequently compared to currently available high end products at several local audio dealers as well as my distributor who carried most of the Kondo stuff. Obviously not every comparison was a win, but it did provide a good reference point, and a very educational experience..
One of the interesting discoveries particularly in shoot outs is that guys with very expensive name brand gear can become very nonplussed when your nasty looking prototype clearly outperforms their very expensive reference. Ego can and does play a large role under these circumstances, and can derail an otherwise successful project. Something to keep in mind when doing this sort of testing is that it is always a good idea to have a small crowd listening at such events, many of whom are not enfranchised and will voice a less biased/more honest opinion of your work
I still have several of the older amplifier prototypes for these units and use them as a frame of reference for some of my newer designs. (not many these days)
In addition I have a lot of opportunity to compare other peoples designs against my own both in my system and in several of theirs. (I'm fortunate to live in an area with widespread and vibrant audio diy.)
Last edited:
Thanks for this reminder Bas. For some unexplicable reason i seem to prefer the sound if a choke is included even if there is additional RC filtering after the choke. So it's not simply "resistors are evil". Not to mention that my R-core mains transformer has surprisingly high DC resistance.
Kevins previous post is probably the best explanation of proper engineering on a tube amp, or anything else where opinions matter. I offer a few of my own observations:
Better to let someone else provide the "known good" stuff. That way you can't be responsible for "rigging the test" with a defective unit. In fact I just bring my amps. Hook them up in place of their amps with everything else the same. Granted you won't always get great results that way, and many people especially young people have neaver learned how to listen.
I have seen that big time, but it can work the other way too. I used to have time to make complete amps for sale. I had a chance meeting with a well to do doctor who was telling me (and several others) about the super high end stereo that he had purchased in Hong Kong. It turned out to be a SE 300B amp that he spent over $5K on. I told him that I had some tube amps and that I built them myself. He said that he was looking for another amp for his bedroom, so I told him that I would bring one next time I was in town. He lived next door to my brother (250 miles away).
I brought a prototype Tubelab SE with 45's mounted in an ugly wood box. My Lexan amp didn't exist yet. Neither did Tubelab Inc. We connected it to his bedroom speakers in place of some sort of small SE tube amp. After a few minutes he wanted to hear it connected to his big horn speakers. I had to leave so I left the amp with him. I was about an hour down the road when my cell phone rang. There were 3 more calls before I got home because he had "discovered" details in his records that he had never heard before. The fact that an ugly plywood box could flat smoke his $5K wonderamp was just not possible.
Well, I made him a nice amp to fit the custom cabinet that he had made, and then another amp, and then THE headphone amp. He drove to my house 3 different times, and spent a few nights staying in a hotel so that I could make him a headphone amp exactly to his liking.
Headphone Amp
In a case like this, you already have the customer, you make him what he wants even if you would never do it that way your self.
Yes, I sold more amps this way than any other way. Both HiFi and guitar amps. In fact there was about a two year period where I couldn't make enough guitar amps. This all started when I loaned my own personal guitar amp to a friend. It ended when he became a crack adict, and stole several peoples amps and guitars!
I work in a facility that used to employ over 1000 engineers. There were about 4 tube heads here, the rest ignored us, or laughed. I had an old P-P KT88 amp that I had built from junk box parts. I loaned it out to anyone who dared to try it. Many of the younger engineers didn't like the sound, because it didn't shake the whole house the way that their multi gigawatt HT box did.
I spend a lot of time with the engineering. Most of my amps start with an idea. The next step may be simulation. Then comes stage by stage prototyping and analysis. During the development process each stage gets subjected to analytical optimization individually, then the entire design gets optimized. At this time the amp has not seen speakers yet. Many designs get scrapped or major modifications at any one of these steps. Once I am happy with the measured performance, the sound testing and further refinement starts. It is possible that the listening tests may force some design changes.
This thread follows a design where an idea was proposed by another forum member, and taken to a proposed amplifier design by two people at opposite ends of the earth without lifting a soldering iron. Then I built the first prototype (of just the driver). It did not work as planned, but some testing led to some circuit modification. After a few more itterations I had a driver that worked perfectly, so I added output stages, tweaked some more, and then hooked up some speakers and let it rip. No further optimization was needed. Several different output tubes were tried, and more will be tried in the near future. I have a working stereo prototype of the amplifier - no power supply yet.
Notice there has been no mention of the power supply, or regulation. All of my amps are designed using bench type regulated supplies. I make the amp work right first using known good power supplies, then I design a power supply as a seperate design excersize.
Often the economics of the design dictate a simple non regulated power supply. There usually audible differences between a good regulated supply and a simple CRC unregulated supply, using quality electrolytic caps. These differences are often not subtle. The obvious issue is hum, but that one is usually easy to fix. The output impedance of the supply will influence the dynamics of the amp (expecially a P-P amp) and the detail (in a SE amp). A low output impedance allows for higher instantaneous peak currents (dynamics) and less intermodulation through the power supply (detail).
Often the hum can be controlled with the addition of a choke, and the output impedance can be lowered by the addition of a quality low ESR capacitor across the supply near the feed to the OPT. I use a good motor run cap for this purpose. I have stated that this is the single most effective low cost mod that you can do to my amp boards, and dozens of customers have agreed. Is this as good as a well designed regulated power supply? Usually not. Is a regulated supply needed? Usually not. In most amplifier designs a regulated supply will offer some measurable and sonic benefit. In many there are other upgrades that are far easier to do that may offer more sonic improvement. These should be addressed first.
This is not always the case. Our power lines have become far dirtier in the last few years. My line power runs 8 to 10% distortion (yes I measure it) where it was 3% before the power company replaced the pole transformer (lightning fried the old one). The B+ voltage in my Simple SE varies by more than 10% over the course of a weekend due to line voltage and line distortion changes. Many amplifiers do not like this (it does not seem to bother the Simple SE). Some amplifiers (like the 6L6GC amp discussed above) use multiple supply rails that interact. This is often the case when multiple stages are directly coupled. These amps often benefit or require a regulated supply. So for these cases, or for the upgrader that wants to regulate the power supply in his pocket watch, the next question is...Tube or SS.
A good regulated power supply is not a trivial design. You want low hum and noise, good load and line regulation, good load and line transient response, and low output impedance from DC to say 100KHz or more. The typical design of a modern linear supply reads like the design of a modern solid state amp. Tons of open loop gain and a stable feedback system. The gain is more easily realized using solid state components, but the pass device may be more reliable if a tube is used. I have done linear regulators that are entirely tube, or entirely SS. The SS regulators measure better. My favorite simple regulator is a variation of the Maida using a big mosfet for the pass device. It will however explode if the output is accidentally shorted during use. I have been experimenting with a hybrid regulator design for the 6L6GC amp, but I have not had much time to work on it. The results will be posted on the 6L6GC thread as it unfolds.
Better to let someone else provide the "known good" stuff. That way you can't be responsible for "rigging the test" with a defective unit. In fact I just bring my amps. Hook them up in place of their amps with everything else the same. Granted you won't always get great results that way, and many people especially young people have neaver learned how to listen.
I have seen that big time, but it can work the other way too. I used to have time to make complete amps for sale. I had a chance meeting with a well to do doctor who was telling me (and several others) about the super high end stereo that he had purchased in Hong Kong. It turned out to be a SE 300B amp that he spent over $5K on. I told him that I had some tube amps and that I built them myself. He said that he was looking for another amp for his bedroom, so I told him that I would bring one next time I was in town. He lived next door to my brother (250 miles away).
I brought a prototype Tubelab SE with 45's mounted in an ugly wood box. My Lexan amp didn't exist yet. Neither did Tubelab Inc. We connected it to his bedroom speakers in place of some sort of small SE tube amp. After a few minutes he wanted to hear it connected to his big horn speakers. I had to leave so I left the amp with him. I was about an hour down the road when my cell phone rang. There were 3 more calls before I got home because he had "discovered" details in his records that he had never heard before. The fact that an ugly plywood box could flat smoke his $5K wonderamp was just not possible.
Well, I made him a nice amp to fit the custom cabinet that he had made, and then another amp, and then THE headphone amp. He drove to my house 3 different times, and spent a few nights staying in a hotel so that I could make him a headphone amp exactly to his liking.
Headphone Amp
In a case like this, you already have the customer, you make him what he wants even if you would never do it that way your self.
Yes, I sold more amps this way than any other way. Both HiFi and guitar amps. In fact there was about a two year period where I couldn't make enough guitar amps. This all started when I loaned my own personal guitar amp to a friend. It ended when he became a crack adict, and stole several peoples amps and guitars!
I work in a facility that used to employ over 1000 engineers. There were about 4 tube heads here, the rest ignored us, or laughed. I had an old P-P KT88 amp that I had built from junk box parts. I loaned it out to anyone who dared to try it. Many of the younger engineers didn't like the sound, because it didn't shake the whole house the way that their multi gigawatt HT box did.
I spend a lot of time with the engineering. Most of my amps start with an idea. The next step may be simulation. Then comes stage by stage prototyping and analysis. During the development process each stage gets subjected to analytical optimization individually, then the entire design gets optimized. At this time the amp has not seen speakers yet. Many designs get scrapped or major modifications at any one of these steps. Once I am happy with the measured performance, the sound testing and further refinement starts. It is possible that the listening tests may force some design changes.
This thread follows a design where an idea was proposed by another forum member, and taken to a proposed amplifier design by two people at opposite ends of the earth without lifting a soldering iron. Then I built the first prototype (of just the driver). It did not work as planned, but some testing led to some circuit modification. After a few more itterations I had a driver that worked perfectly, so I added output stages, tweaked some more, and then hooked up some speakers and let it rip. No further optimization was needed. Several different output tubes were tried, and more will be tried in the near future. I have a working stereo prototype of the amplifier - no power supply yet.
Notice there has been no mention of the power supply, or regulation. All of my amps are designed using bench type regulated supplies. I make the amp work right first using known good power supplies, then I design a power supply as a seperate design excersize.
Often the economics of the design dictate a simple non regulated power supply. There usually audible differences between a good regulated supply and a simple CRC unregulated supply, using quality electrolytic caps. These differences are often not subtle. The obvious issue is hum, but that one is usually easy to fix. The output impedance of the supply will influence the dynamics of the amp (expecially a P-P amp) and the detail (in a SE amp). A low output impedance allows for higher instantaneous peak currents (dynamics) and less intermodulation through the power supply (detail).
Often the hum can be controlled with the addition of a choke, and the output impedance can be lowered by the addition of a quality low ESR capacitor across the supply near the feed to the OPT. I use a good motor run cap for this purpose. I have stated that this is the single most effective low cost mod that you can do to my amp boards, and dozens of customers have agreed. Is this as good as a well designed regulated power supply? Usually not. Is a regulated supply needed? Usually not. In most amplifier designs a regulated supply will offer some measurable and sonic benefit. In many there are other upgrades that are far easier to do that may offer more sonic improvement. These should be addressed first.
This is not always the case. Our power lines have become far dirtier in the last few years. My line power runs 8 to 10% distortion (yes I measure it) where it was 3% before the power company replaced the pole transformer (lightning fried the old one). The B+ voltage in my Simple SE varies by more than 10% over the course of a weekend due to line voltage and line distortion changes. Many amplifiers do not like this (it does not seem to bother the Simple SE). Some amplifiers (like the 6L6GC amp discussed above) use multiple supply rails that interact. This is often the case when multiple stages are directly coupled. These amps often benefit or require a regulated supply. So for these cases, or for the upgrader that wants to regulate the power supply in his pocket watch, the next question is...Tube or SS.
A good regulated power supply is not a trivial design. You want low hum and noise, good load and line regulation, good load and line transient response, and low output impedance from DC to say 100KHz or more. The typical design of a modern linear supply reads like the design of a modern solid state amp. Tons of open loop gain and a stable feedback system. The gain is more easily realized using solid state components, but the pass device may be more reliable if a tube is used. I have done linear regulators that are entirely tube, or entirely SS. The SS regulators measure better. My favorite simple regulator is a variation of the Maida using a big mosfet for the pass device. It will however explode if the output is accidentally shorted during use. I have been experimenting with a hybrid regulator design for the 6L6GC amp, but I have not had much time to work on it. The results will be posted on the 6L6GC thread as it unfolds.
At some point, most DIYers will start using their previous amps as reference. Most gets a favorite at some point.
The game about comparing to high dollar commercial gear, gets rather old.
Magura
Oh, this is a feedback loop, so the plate voltage of the error amplifier is self-regulating. Assuming B+ (?) is the output voltage, the requlator will always maintain the plate voltage of the pentode at B+ minus the gate-to-cathode voltage of the series pass tube.
Don't forget to probe for HF oscillation.
Quite right, one other thing I will mention is make sure that the open loop bandwidth of the error amplifier with the CCS is at least 20kHz which probably means adding some resistance to define the load impedance. In theory it should make no difference to the sound, but in a number of listening tests "victims" preferred the higher bandwidth, The required resistance can go right across the CCS and could probably be several hundred K ohms, This may sound counter-intuitive but the CCS will give you the required Ip for higher transconductance resulting in higher gain than the current that the load resistor would flow on its own.
GK, Kevin,
I just finished testing the V-Reg with the CCS
The results are amazing; you can hear the difference, no question about it.
The sound is sharper and more detailed than ever, and I'm setup in the basement on the cheap speaker!
Here are some of the details, I'll upload the schematic tomorrow.
3mA CCS for the plate of the pentode
(so now the pentode is operating at the “typical” value)
20mA CCS for the 0A3
With some on my earlier experiments, the 0A3 would flicker at 20Hz.
The 20mA CCS took care of that!
This setup allows the B+ to range from 258V to 360V
And the Z-out…….. .15R at 20Hz
Thank You
Having just read Floyd Toole's Sound Reproduction, i like what he says... the measurements are only valid if they can be correlated with what 2 ears + a braln perceive (unbiased, ie blind). Unlike him i believe that not blind results by trained listeners can be useful.
dave
Absolutely Dave. While I trust the engineered circuit to tell me that it works as designed, the final say as to whether it "sounds good" is up to my two ears and my brain
Actually, it's sub-100 milliohms Zout and sub-300 uV hum/noise
http://www.linearaudio.nl/t-reg-1.htm
jd
It is very exelent! And easy to make it!
Thank
Attachments
Question for Janneman
Nice work on your tube regulator and website! I am, however, left with one question: What circuit and technique do you use for measuring output impedance with the AP system? Would you happen to have that described anywhere?
Thanks,
~Tom
Actually, it's sub-100 milliohms Zout and sub-300 uV hum/noise
http://www.linearaudio.nl/t-reg-1.htm
jd
Nice work on your tube regulator and website! I am, however, left with one question: What circuit and technique do you use for measuring output impedance with the AP system? Would you happen to have that described anywhere?
Thanks,
~Tom
If killing the loop gain (and just about every performance parameter) of the regulator really did (rather then just imagined) make the amplifier sound better, then the regulator most probably was a primitive design having inadequate frequency compensation and conditional / marginal stability.
If killing the loop gain (and just about every performance parameter) of the regulator really did (rather then just imagined) make the amplifier sound better, then the regulator most probably was a primitive design having inadequate frequency compensation and conditional / marginal stability.
The changes in question at most reduce the open loop gain by just a few dB, and should be sufficient to ensure constant (if somewhat lower) gain out to at least 20kHz. Solid state amplifier designers have reported similar findings in the design of VAS in the past. (Walter Jung, Matti Otala amongst others.)
I was skeptical too, and they were talking specifically about one of my designs - apparently a primitive design* having inadequate frequency compensation and condition/marginal stability although there was absolutely no evidence to support your contentions based on extensive bench measurements and observation..
Some well grounded engineering types messing with my designs claimed it did make a difference so I listened. The difference was subtle enough that I doubt double blind testing would result in a clear conclusion, however as a design criteria it is free, and might be audible under some circumstances in amplifier designs with poor psrr. *Yes it actually was quite primitive having just a simple cascode error amp resistively loaded, (with a deliberately bad tube choice for availability reasons at that time) a boot-strapped voltage reference, and a pentode pass element. It was unconditionally stable, and had good if not particularly noteworthy performance by the standards of anything other than tube audio. Line/load regulation of a couple of % or better, output noise broadband of <1mVpp, source impedances down to a couple of ohms in some cases..(Line: +/- 20% of 120V nominal, Load: 0 - 100% rated current)
Can you post an example of how you would implement a relatively high performance all vacuum tube regulator?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.