The slope of the line that is tangent[...]
Bias in tape recorders is used to overcome effect of hysteresis to magnetize magnet layer of the tape.
An externally hosted image should be here but it was not working when we last tested it.
Bias in tape recorders is used to overcome effect of hysteresis to magnetize magnet layer of the tape.
Joe Rasmussen said:
Just wanted to add re the above graph: The plateauing is dependant on the speaker's voltage sensitivity. The higher +X dB the sensitivity, the plateau will go down -X dB. So if you use high voltage sensitivity (not really related to nominal speaker Z), then you need the bias more. So in the example shown above -70dB is actually 50dBSPL. But with most speakers it will be around 60-65dBSPL and hence will be able to deal with even super sensitive horn speakers.
So now I have revealed how the numbers stack up. With most conventional systems with dome tweeters, the VC will not exceed 10mW added dissipation.
Joe R.
"Bias in tape recorders is used to overcome effect of hysteresis to magnetize magnet layer of the tape."
Wavebourn makes a good point, the recording process is concerned with the absolute permanent magnetic record left on the tape, so eliminating hysteresis was important there. But transformers operate by change in flux producing a voltage (not absolute flux).
So if you drive a transformer from a low Z source, voltage leads to flux change and flux change leads to induced secondary voltage. Hysteresis is invisible as far as the main signal is concerned. Hysteresis only shows up as weird magnetizing current (which could cause some distortion if the tube is not sufficiently in control, ie, a high Z output stage). So for high Z out applications, I could see some possible benefit of the HF bias.
Then there is the pin-striped xfmr scheme, with high Mu permalloy lams added to the stack to handle small signals. Seems like that should be helpful, although I haven't seen any actual measured data to support it.
Don
Wavebourn makes a good point, the recording process is concerned with the absolute permanent magnetic record left on the tape, so eliminating hysteresis was important there. But transformers operate by change in flux producing a voltage (not absolute flux).
So if you drive a transformer from a low Z source, voltage leads to flux change and flux change leads to induced secondary voltage. Hysteresis is invisible as far as the main signal is concerned. Hysteresis only shows up as weird magnetizing current (which could cause some distortion if the tube is not sufficiently in control, ie, a high Z output stage). So for high Z out applications, I could see some possible benefit of the HF bias.
Then there is the pin-striped xfmr scheme, with high Mu permalloy lams added to the stack to handle small signals. Seems like that should be helpful, although I haven't seen any actual measured data to support it.
Don
This thread started with Xenu asking the question of any "modern" amplifier designs using tubes - a fair queston, since nearly all commercially available tube amplifiers are basically minor variations of 1950's "Golden Age" designs, with the occasional SET for entertainment value.
I can think of two DIY amplifiers that are significantly different than anything on the market: Gary Pimm's Tabor amplifier, and Steve Bench's modern feedforward version of the Western Electric Harmonic Balancer.
The Gary Pimm family of amplifiers (he's built a number of variations) are balanced transconductance amplifiers - a steered current source, if you will, and works best not with triodes, but high-Z pentodes, or even MOSFETs. They do NOT work if you try to "improve" them by converting to triode - the higher the Zout of the device, the better they work (lower distortion, more bandwidth). You have to read Gary's page very carefully to understand the current-mode operation of the amplifier - it goes over the heads of most people who read it.
The Steve Bench revival of the Harmonic Balancer (which does not appear on his web-page, but he has built and measured with spectacular results) is a technique first discovered by Bell Labs that wasn't global feedback, but a method of canceling odd-order harmonic terms in a balanced amplifier. Since the dominant form of distortion in a balanced amplifier is odd-order (3rd, 5th, 7th, 9th, et al), and there is no impact on HF phase margin (unlike feedback), this is a significant advancement that was unjustly forgotten for nearly seventy years. Steve's technique uses an independent stage that injects a careful proportion of the HB signal to null out the odd-order harmonics (WE used a resistor bridge inside the amplifier).
Neither sounds anything like commercial high-end amplifiers, and the operating principles are quite different than the usual Williamson, Dyna, Acro, Marantz, or McIntosh variant. To the best of my knowledge, there are no commercial examples, although it is possible that Gary or Steve might commercialize them, because the inherent linearity of these circuits is several times better than the usual PP-with-feedback amplifier - and as a result, they sound different, as well.
It's been my experience that the best (lowest intrinsic distortion) amplifiers don't "converge" towards transistor sound - instead, they reveal Class AB transistor amplifiers as having an odd low-resolution "digital" kind of coloration of their own, something akin to MP3 sound. The highest resolution amplifiers also reveal the most about the source - the subtle differences between 30 IPS mastertapes and 96/24 digital, for example.
In the pro universe, studio monitors and amplifiers are rank-ordered by their ability to distinguish between 16, 20, and 24-bit (and analog) sources. Using this subjective yardstick, much of commercially available high-end equipment would not pass muster. The high-quality DIY projects will - I've heard the results for myself.
Here's some related reading on these topics: Current Loops and the Amp-Speaker Interface, and some unusual aspects of Western Electric circuits. Have fun!
I can think of two DIY amplifiers that are significantly different than anything on the market: Gary Pimm's Tabor amplifier, and Steve Bench's modern feedforward version of the Western Electric Harmonic Balancer.
The Gary Pimm family of amplifiers (he's built a number of variations) are balanced transconductance amplifiers - a steered current source, if you will, and works best not with triodes, but high-Z pentodes, or even MOSFETs. They do NOT work if you try to "improve" them by converting to triode - the higher the Zout of the device, the better they work (lower distortion, more bandwidth). You have to read Gary's page very carefully to understand the current-mode operation of the amplifier - it goes over the heads of most people who read it.
The Steve Bench revival of the Harmonic Balancer (which does not appear on his web-page, but he has built and measured with spectacular results) is a technique first discovered by Bell Labs that wasn't global feedback, but a method of canceling odd-order harmonic terms in a balanced amplifier. Since the dominant form of distortion in a balanced amplifier is odd-order (3rd, 5th, 7th, 9th, et al), and there is no impact on HF phase margin (unlike feedback), this is a significant advancement that was unjustly forgotten for nearly seventy years. Steve's technique uses an independent stage that injects a careful proportion of the HB signal to null out the odd-order harmonics (WE used a resistor bridge inside the amplifier).
Neither sounds anything like commercial high-end amplifiers, and the operating principles are quite different than the usual Williamson, Dyna, Acro, Marantz, or McIntosh variant. To the best of my knowledge, there are no commercial examples, although it is possible that Gary or Steve might commercialize them, because the inherent linearity of these circuits is several times better than the usual PP-with-feedback amplifier - and as a result, they sound different, as well.
It's been my experience that the best (lowest intrinsic distortion) amplifiers don't "converge" towards transistor sound - instead, they reveal Class AB transistor amplifiers as having an odd low-resolution "digital" kind of coloration of their own, something akin to MP3 sound. The highest resolution amplifiers also reveal the most about the source - the subtle differences between 30 IPS mastertapes and 96/24 digital, for example.
In the pro universe, studio monitors and amplifiers are rank-ordered by their ability to distinguish between 16, 20, and 24-bit (and analog) sources. Using this subjective yardstick, much of commercially available high-end equipment would not pass muster. The high-quality DIY projects will - I've heard the results for myself.
Here's some related reading on these topics: Current Loops and the Amp-Speaker Interface, and some unusual aspects of Western Electric circuits. Have fun!
Could the tubelabe SE board have the outputs after the Output Trans. be connected so the 2 Channels are connected together?
I need the power for my old speakers! Would one get 14-18 W/CH with 300B tubes using two separated boards and PWR Trans? Should sound better with everything totally separate!
How would 845's work instead of the 300B's running at the 300B's voltage or a bit higher? Should last forever? 845's are only $45ea! Maybe get similar power as 300B's 45's seem to be hard to get and expensive now. I want to try DHT SE.
Thanks Very Much!
Randy
I need the power for my old speakers! Would one get 14-18 W/CH with 300B tubes using two separated boards and PWR Trans? Should sound better with everything totally separate!
How would 845's work instead of the 300B's running at the 300B's voltage or a bit higher? Should last forever? 845's are only $45ea! Maybe get similar power as 300B's 45's seem to be hard to get and expensive now. I want to try DHT SE.
Thanks Very Much!
Randy
I must agree. This thread has given me a few new ideas and a few things were discussed that I will need to try for myself. I don't see a whole lot of experiment time in the near future though.
I believe so, but I haven't tried it. If you parallel the two channels after the OPT's you need to use half of the rated load impedance. In other words use an 8 ohm load on two parallelled 16 ohm windings. You can series two 4 ohm windings with an 8 ohm load.
I have paralleled the two channels BEFORE the OPT's on the Tubelab SE with good results. It is possible and desirable to pull one of the driver tubes, using one 5842 to drive both output tubes. Lift the plate end of one coupling cap and run it over to the plate of the remaining 5842. The mosfet buffers are a very light load so the 5842 has no problem driving two of them. Use an OPT with half of the desired load impedance. 1500 to 2500 ohms instead of 3000 to 5000 ohms. This allows a seperate bias adjustment for each tube. One user has tried paralleling externally mounted tubes but this requires carefully matched tubes since there is only one bias adjustment.
I have not tried the 845 on anything lower than 900 volts. I have heard of someone running them on 600 volts, but nothing lower. The 845 is a hard tube to drive and usually requires more than one driver stage to get enough gain.
Lynn Olson said:
Lynn, two points on this.
Didn't Western Electric refer to their circuit as the 'harmonic equalizer?"
Just trying to keep the terminology straight on that.
Point two. Would you consider the Steve Bench matrix amplifier to be
an unusual modern design? I know that the matrix concept was dabbled with briefly in the '50s. Steve was enthusiastic about his 813 matrix, although his WEHE amp may have superceded that place in his estimation.
Gary
OK, I promised earlier on that I would post a couple more [conceptual] schematics, so here is one. Please note that there is a bit more to the eye and yes, those are input Fets and yes, there is a good reason why. That input stage also has an open secret and I wonder who is going to pick, like "where have I seen that?" And if you figure that out, yes, I had a hand in that too (that's a hint). I will give another hint, it is used in the most other extreme part of the audio chain.
The name "Forced Symmetry" was coined in the 90's and symmetry became somewhat of a buzz word after that. In that case, just looking at the diagram makes the description, rather than the phrase, obvious. But here symmetry was kinda by extension inspired by Hedge who, IMMSMR, was about about improving amplifiers using less than perfect output transformers - add long tail etc, but of course here the idea is to make the amp as symmetrical as possible notwithstanding using poor parts.
Any comments:
BTW, I built a number of these between '90s.
Could this be given the WEHE treatment? Probably for those interested.
Joe R.
The name "Forced Symmetry" was coined in the 90's and symmetry became somewhat of a buzz word after that. In that case, just looking at the diagram makes the description, rather than the phrase, obvious. But here symmetry was kinda by extension inspired by Hedge who, IMMSMR, was about about improving amplifiers using less than perfect output transformers - add long tail etc, but of course here the idea is to make the amp as symmetrical as possible notwithstanding using poor parts.
Any comments:
An externally hosted image should be here but it was not working when we last tested it.
BTW, I built a number of these between '90s.
Could this be given the WEHE treatment? Probably for those interested.
Joe R.
I've seen the two local loop deal before- I think it was either in an RCA tube manual circuit (the 7027 amp) or a Berning amp- but not often and not with a CC FET cascode diff amp. The latter is something I tried, with a 2N5566 on the bottom and ECC88 on top. Worked OK, but stabilizing a very high feedback amp like that is not trivial!
SY said:
I wouldn't mind seeing that circuit.
I would not use a heap of a lot of feedback. But the Anode-to-Anode (AA) loop pair was tried on its own and while it sort of worked it was also problematic in a different way. There was a great sense of clarity but also tended to sound lightish in balance. The the Anode-to-Source (AS) loop pair was introduced and this had the exact opposite effect. If AS was balanced against AA, then the balance could be "dialled" in. If AS was allowed to dominate, then the sound would sound to rich and too dark and taking out AA alltogether and it would sound positively weird.
This amp is peculiar as you can tonally alter/tune its sound in a way I have never experienced before.
Bandersnatch said:hey-Hey!!!,
Try the triode-on-the-bottom FET/triode cascode. Alledgedly it is not as quiet, but it makes a fine pentode-ish construct...
I did of course first use all-triode (no fets) but it made the feedback loop way too low impedance (and a lot of heat). The higher transconductance of the fet means the two source resistors' value can become many times higher and this makes the resistances in the loops also many times higher and much less DC current shunted back. Don't forget that all that current, the input stage plus the four loops, all has to be pulled through the fet CCS.
Also, a nice byproduct, the fets are matched and stay that way long term. Using a tube here the symmetry will not be stable long term, even assuming you could find tubes that are matched. This is 2SK147/170/369 and the the tube on top also has to have fairly high (but not as much as the fets) gm. Otherwise the Cathode will swing too much and the fet will be 'starved' of voltage. This may have to swing several hundred volts on the Anode and yet the Cathode be as stable as possible. I chose the Sovtek 6922 as it has higher rated voltages than standard ECC88/6DJ8.
But this fet/tube/casode stage used before and since the 80's?
Joe R.
Joe Rasmussen said:
hey Joe,
I have been running 6H6pi/FQP1N60 front end since 2004. the amps have had a lot of time on them, and the plate-plate( errr...drain-drain ) voltage has not drifted more than 10V on the original pair of tubes. The tube wear thing is trivial IMO, at least with those tubes.
With a FET under the triodes, and facing the issue of 'starving' the FET, I think you're getting only part of the cascode benefit. If the d-s voltage varies like you describe, the g-d capacitance will be making itself visible. OTOH, the FET above the triode will deliver a near-vertical load for the triode and its plate-cathode voltage will be very close to constant...which is what we're after.
cheers,
Douglas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.