Hello,
today I received my copy of Merlin Blencowe's book Designing High Fidelity Tube preamps. I opened it about in the middle and fell into chapter 6.1.3: Cascode BJT Current Source. Merlin writes that a CCS's dynamic impedance can be increased dramatically by paralleling the reference LED's or zeners with a capacitor. But it takes absurd high capacitance values to keep the impedance high in the LF region also. Now there are gold capacitors with values in the F(arad) region. Were they suitable for this purpose if the total reference voltage is kept well beneath their voltage rating?
Best regards!
today I received my copy of Merlin Blencowe's book Designing High Fidelity Tube preamps. I opened it about in the middle and fell into chapter 6.1.3: Cascode BJT Current Source. Merlin writes that a CCS's dynamic impedance can be increased dramatically by paralleling the reference LED's or zeners with a capacitor. But it takes absurd high capacitance values to keep the impedance high in the LF region also. Now there are gold capacitors with values in the F(arad) region. Were they suitable for this purpose if the total reference voltage is kept well beneath their voltage rating?
Best regards!
Besides the issue jean-paul brought up, you have to ensure supercapacitors remain cool. Their lifetime is not very long and it gets worse rapidly with increasing temperature. There are also big differences from one brand to another.
Could you post a sketch of the schematic you have in mind? I think it would help people to understand what you mean.
Could you post a sketch of the schematic you have in mind? I think it would help people to understand what you mean.
If used wrongly they are nice hand grenades too.
BTW I think you refer to supercapacitors (like MarcelvdG also pointed out) and not to the older gold caps. Gold caps were those parts that needed to be replaced after years in for instance FM tuners. These had sulfuric acid in them and had somewhat lead-acid battery like properties.
BTW I think you refer to supercapacitors (like MarcelvdG also pointed out) and not to the older gold caps. Gold caps were those parts that needed to be replaced after years in for instance FM tuners. These had sulfuric acid in them and had somewhat lead-acid battery like properties.
Last edited:
Yes Panasonic called their EDLCs Gold caps but that became their family name. You will know Luxaflex I think 
There are many variants but todays li-ion super/ultracapacitors are different beasts from the older memory keeping relatively high ESR gold capacitors. They are even used to store energy coming from braking with e-cars (shunting the electromotors) and are able to deliver very high currents. Some versions are more battery than capacitor.
Some use these (wisely) for spot welding.
I see them turning up in power electronics and am a little concerned with their risks. Same goes for Lithium Polymer batteries in power electronics. Like many new developments they become inescapable buzzwords and things are seriously pushed aggressively by their manufacturers. Low weight, high energy density and small size are instant winners today. Then they are tested/implemented but when stuff goes wrong it seriously goes wrong.
There are many variants but todays li-ion super/ultracapacitors are different beasts from the older memory keeping relatively high ESR gold capacitors. They are even used to store energy coming from braking with e-cars (shunting the electromotors) and are able to deliver very high currents. Some versions are more battery than capacitor.Some use these (wisely) for spot welding.
I see them turning up in power electronics and am a little concerned with their risks. Same goes for Lithium Polymer batteries in power electronics. Like many new developments they become inescapable buzzwords and things are seriously pushed aggressively by their manufacturers. Low weight, high energy density and small size are instant winners today. Then they are tested/implemented but when stuff goes wrong it seriously goes wrong.
Last edited:
I see. You have a point that you need a reasonably low ESR for Kay's application.
I designed and built a power electronic circuit with 300 F supercapacitors a couple of years ago. It had a control loop that was largely in firmware, so the slightest programming error could result in overvoltage. I monitored the voltages closely and nothing ever got damaged, but still, reading your hand grenade remark, I'm glad I used a 100 times scaled down version with small 3 F capacitors for debugging.
I designed and built a power electronic circuit with 300 F supercapacitors a couple of years ago. It had a control loop that was largely in firmware, so the slightest programming error could result in overvoltage. I monitored the voltages closely and nothing ever got damaged, but still, reading your hand grenade remark, I'm glad I used a 100 times scaled down version with small 3 F capacitors for debugging.
It's basically the usual cascode CCS schematics. Merlin B. added a - big - capacitor in parallel with the reference diodes, i. e. from ground to the upper transistor base and found out that it, in comparison with the original, capacitor-less design, dramatically increases the CCS impedance in the 1 kHz to 4 kHz region. But he didn't state why, where from the increase results. It can actually be assumed that a low ESR (in comparison with the dynamic impedance of LED's) helps, but I dunno.
Anyway, Jean-Paul is right. Charging a, say, 1 F super capacitor (or Gold Cap) from 0 V to 3,6 V (the voltage drop across two red LED's in MB's example) by, say, 5 mA takes a time of about 720 seconds or 12 minutes. Far from one hour, but absolutely inacceptable. Hence, I'd better discard my idea.
Btw, Merlin mentions 0.18 F (charging time 130 s @ 5 mA), but ruled out a capacitor of this value for practical reasons.
Best regards!
Anyway, Jean-Paul is right. Charging a, say, 1 F super capacitor (or Gold Cap) from 0 V to 3,6 V (the voltage drop across two red LED's in MB's example) by, say, 5 mA takes a time of about 720 seconds or 12 minutes. Far from one hour, but absolutely inacceptable. Hence, I'd better discard my idea
.Btw, Merlin mentions 0.18 F (charging time 130 s @ 5 mA), but ruled out a capacitor of this value for practical reasons.
Best regards!
Adding a capacitor in parallel with reference diode sounds to me like: lowering the impedance of the reference voltage. Which you could accomplish just as well with a few solid state devices and a smidgen of circuit savvy -- and it'd stabilize to equilibrium in milliseconds not minutes.
But do you really want solid state devices in a bias circuit? (apart from reference diodes which are of course solid state)? Would it offend all that is holy and righteous, to build a tube amp and employ solid state devices having more than two terminals? Even though the solid state parts are definitely not in the signal path?
But do you really want solid state devices in a bias circuit? (apart from reference diodes which are of course solid state)? Would it offend all that is holy and righteous, to build a tube amp and employ solid state devices having more than two terminals? Even though the solid state parts are definitely not in the signal path?
Not even that. Sometimes a red LED mounted inmidst the pin circle just to illuminate a cold tube suffices to impress clueless audiophools with the quality of this tube amp, while a plain vanilla LM3886 or the like does the job beneath the chassis .
Although I'm very fascinated of tube amplification, I'm far from the »No SS at all!« claim. And I also won't say that a SS CCS as a LTP's tail won't be noticeable at all. Quite the contrary: Improving the performance of this stage surely will also improve the amplifier's sonic performance.
I've always been fascinated by the efforts others have done in the past to improve the performance of tube amplifiers, say Hafler, Kereos, Gow, McIntosh, Nestotovic, just to name a few. I'm convinced that if they'd had these cheap SS devices then, they'd also used them with enthusiasm.
Best regards!
.Although I'm very fascinated of tube amplification, I'm far from the »No SS at all!« claim. And I also won't say that a SS CCS as a LTP's tail won't be noticeable at all. Quite the contrary: Improving the performance of this stage surely will also improve the amplifier's sonic performance.
I've always been fascinated by the efforts others have done in the past to improve the performance of tube amplifiers, say Hafler, Kereos, Gow, McIntosh, Nestotovic, just to name a few. I'm convinced that if they'd had these cheap SS devices then, they'd also used them with enthusiasm.
Best regards!
I tried some (at the time quite expensive) constant current diodes (j-fet based stuff i guess) from motorola some decades ago but quickly abonded them and went back to tubes (plain resistors where even better in many respects) because of the quite low impedance at the upper end of the audio band. Additionally, the tempco made those ccs-diodes more usuable as temp sensors than ccs. I cannot remember the details other than that they where a total disappointment and expensive mistake. At the same time i tried some 100V j-fet from TI that turned out quite usable otherwise in my first hybrid, especially as phase splitter.
Does anyone have info about the realistic qualties of nowdays semiconductor based ccs circuits at high voltage, high frequency swings compared to a pentode ccs?
Does anyone have info about the realistic qualties of nowdays semiconductor based ccs circuits at high voltage, high frequency swings compared to a pentode ccs?
Last edited:
Post #12 doesn't propose a Constant Current Source made of semiconductors.
Rather, it proposes a standard Merlin Blencowe vacuum tube Constant Current Source, except replace the voltage reference diodes and multi-Farad Supercapacitors with an ultra low output impedance voltage reference (actually, voltage regulator) circuit. This ultralow output impedance voltage reference contains 6 or fewer silicon transistors (stripped down versions with 3 devices are possible, for those who mistrust solid state), and is not in the signal path. The only thing it does, is provide VREF to Merlin Blencowe's vacuum tube CCS.
Rather, it proposes a standard Merlin Blencowe vacuum tube Constant Current Source, except replace the voltage reference diodes and multi-Farad Supercapacitors with an ultra low output impedance voltage reference (actually, voltage regulator) circuit. This ultralow output impedance voltage reference contains 6 or fewer silicon transistors (stripped down versions with 3 devices are possible, for those who mistrust solid state), and is not in the signal path. The only thing it does, is provide VREF to Merlin Blencowe's vacuum tube CCS.
Tnx for clearing that up Mark.
Offcourse, if there would have been posted a CD it would have been clear from the beginning.
Luckely, I still have left some excellent 2 legged voltage reference circuits (TAA550) for this purpose.
To me it look like supercaps have nothing to offer in this application (esr to high)
Offcourse, if there would have been posted a CD it would have been clear from the beginning.
Luckely, I still have left some excellent 2 legged voltage reference circuits (TAA550) for this purpose.
To me it look like supercaps have nothing to offer in this application (esr to high)
Last edited: