I am quite happy to send you the top and bottom layers in monochrome PDF so you can etch your own, if i understand you correctly?
But honestly you cannot get the base material FR4 for the price you pay the Chinese for a professional grade ENIG-ROHS board. The link to the group buy is in the first post.
This circuit is very much overkill for ECC88 and the likes, your better of using a LD1085 or LM317 regulator for those, there is very little to gain from a high impedance current source on indirectly heated tubes.
But honestly you cannot get the base material FR4 for the price you pay the Chinese for a professional grade ENIG-ROHS board. The link to the group buy is in the first post.
This circuit is very much overkill for ECC88 and the likes, your better of using a LD1085 or LM317 regulator for those, there is very little to gain from a high impedance current source on indirectly heated tubes.
Hi,
I’m sure there is room for more than one heater regulator, whether it’s yours, Coleman’s or Tentlabs’s. Just keep doing what you are doing (others will probably do the same).
Personally I do appreciate all the gerber files you have posted for all sorts of circuits. I wish there was a forum entry just for public diy audio gerbers. You sure would deserve a place there!
Regards, Gerrit
I’m sure there is room for more than one heater regulator, whether it’s yours, Coleman’s or Tentlabs’s. Just keep doing what you are doing (others will probably do the same).
Personally I do appreciate all the gerber files you have posted for all sorts of circuits. I wish there was a forum entry just for public diy audio gerbers. You sure would deserve a place there!
Regards, Gerrit
About that transistor plaything.
If i had to make my own transistor plaything, it wouldn't be too hard.
Skizlai pair capacitor multiplier on the top and a resistor on the output with a capacitor back to the capacitor multiplier to AC couple the ripple back into the multiplier in reversed phase.
Then a simple two transistor current source on the bottom end, and a NPN VBE multiplier in the middle to make sure both transistors see approximately the same voltage and dissipation. You may need to smack in a RC network to keep the thing from going unstable but thats about it.
But Whats the fun in something that you know from tube building literature wears out the emissive surface on the filaments prematurely?
Also, I want something that i dont have to adjust every time i change a pair of output tubes.
Furthermore, ALL DHT* tubes are specified for a nominal VOLTAGE not CURRENT. You can read much more on that on the Emission labs website. Or any data sheet for that matter.
*There are a handfull of exceptions of types the manufacturer also specified for series connected use and hence have a current rating as well.
If i had to make my own transistor plaything, it wouldn't be too hard.
Skizlai pair capacitor multiplier on the top and a resistor on the output with a capacitor back to the capacitor multiplier to AC couple the ripple back into the multiplier in reversed phase.
Then a simple two transistor current source on the bottom end, and a NPN VBE multiplier in the middle to make sure both transistors see approximately the same voltage and dissipation. You may need to smack in a RC network to keep the thing from going unstable but thats about it.
But Whats the fun in something that you know from tube building literature wears out the emissive surface on the filaments prematurely?
Also, I want something that i dont have to adjust every time i change a pair of output tubes.
Furthermore, ALL DHT* tubes are specified for a nominal VOLTAGE not CURRENT. You can read much more on that on the Emission labs website. Or any data sheet for that matter.
*There are a handfull of exceptions of types the manufacturer also specified for series connected use and hence have a current rating as well.
Thanks, Il see about posting the EAGLE files for this design so people can improve upon it, it was my hobby project for a while because sometimes i get obsessed about a particular thing.
I didnt deserve the response i got from him, my circuit is good, as good as tents but with CURRENT LIMIT. that simple 3 cent diode coupled with the 680mOhm resistors makes sure that you can short the supply indefinitely (@2A output setting and maximum 7.5V voltage setting)
Edit in all fairness a transistor plaything will also do the same.
The tents use a much lower value of shunt resistor, this means that the noise/error signal at the source of the FET will be too small for the opamp to regulate away. Tents circuit uses a MOSFET that has such a high gate capacitance that a normal non buffered opamp cannot swing enough to get rid of the last bit of ripple.
I would like to point out that tent uses this lower value, so you can get the circuit to go from 5VAC to 5VDC. If you eat 500mV in the shunt alone you can't get 5VDC from 5VAC. And that product is aimed at the UPGRADE market, not the BUILDERS market.
Furthermore early versions of the tent didnt have the capacitor over the opamp, I have heard people that bought second hand units, that where oscillating with 1 meter of wire because of the inductance of said wires. I noticed this during testing when i had 2 meters of wire between the regulator and load resistor and spend the better part of a week trying different RC filter networks and capacitors to quiet the oscillation and maintain good output ripple at the same time. This took me so long, because its hard to model all the factors into a spice simulation.
Last edited:
I have some of your DHT heaters for my upcoming GM70 build and I haven't managed to blow them up yet! I hope I soon get a chance to bring them to work with me where I have access to better stuff than my (t)rusty stack of Fluke 8842's (they haven't been calibrated since a long time so I only believe them for about 4 1/2digits)
Don't worry and keep on trucking
Don't worry and keep on trucking
Last edited:
I have some of your DHT heaters for my upcoming GM70 build and I haven't managed to blow them up yet! I hope I soon get a chance to bring them to work with me where I have access to better stuff than my (t)rusty stack of Fluke 8842's (they haven't been calibrated since a long time so I only believe them for about 4 1/2digits)
Don't worry and keep on trucking
Perhaps you shouldnt work so hard man, and finish that PHD
)
I have a version for use with those tubes, someone else paid for the development and a batch of PCB's so i think it is fair i ask if i can post the design here. It is essentially a TENT with a Monster Linear FET. IXTH80N075L2 and a big-*** low inductance TO247 shunt resistor, That shunt was just to get two cool looking 247 devices on the PCB.
Maximum current is about 15A with this version, i have tested it at 8A and it gave about 800uV of noise+hum. That is because the LM7323MAX i selected for the CCS driver does not have a really good small-signal gain. I had trouble finding a part with enough gain with just a few uV error signal from the shunt to drive that monster 20nF gate capacitance of two parallel IXTH80N075L2 i fixed that problem in the MONSTERREG you can find in the free gerbers thread but i used the LH0002 buffer there as a sledgehammer solution.
I tried selling a pair on DIYAUDIO swapmeet but found no takers. Shame because if you cooled it right, you could short it out with a screwdriver at 10V 10A and the fets would just shrug it off like its noting...
Maximum current is about 15A with this version, i have tested it at 8A and it gave about 800uV of noise+hum. That is because the LM7323MAX i selected for the CCS driver does not have a really good small-signal gain. I had trouble finding a part with enough gain with just a few uV error signal from the shunt to drive that monster 20nF gate capacitance of two parallel IXTH80N075L2 i fixed that problem in the MONSTERREG you can find in the free gerbers thread but i used the LH0002 buffer there as a sledgehammer solution.
I tried selling a pair on DIYAUDIO swapmeet but found no takers. Shame because if you cooled it right, you could short it out with a screwdriver at 10V 10A and the fets would just shrug it off like its noting...
Last edited:
Wether you get 1mV or 5mV of noise is unimportant, wether the CCS is ultra high impedance across the audio band, is unimportant. However the real kicker is in the soft start capability. If you switch on a GM70 for example the cold resistance means the filaments will have a resistance that is 1:5 of the warm resistance. (Source Tempco chart on pure tungsten @2700K VS 300K) If you build a voltage driven current source that starts at 150% of nominal operating current, you have an ideal product from a tube filament lifetime point of view. The near zero noise on your fancy handcrafted horns is just an added bonus.
If you AC heat any tube like the GM70 or most of the big transmitter glass for that matter, You lose the ability to avoid temperature shock, One simple trick is to put a NTC in series with the primary, but that can be hard to engineer so it provides the same functionality as something i put online.
If you AC heat any tube like the GM70 or most of the big transmitter glass for that matter, You lose the ability to avoid temperature shock, One simple trick is to put a NTC in series with the primary, but that can be hard to engineer so it provides the same functionality as something i put online.
Last edited:
How were regulated heater supplies done in the 80s? much difference with the current state of the art or merely less fancy components available
Member
Joined 2009
Paid Member
Quick off topic: I’m also interested in r.f. heating. I am thinking of trying one of those ‘electronic transformers’ used for halogen lights to heat up an 845
You don’t have that Glass Audio article as a link?
p.s. always tends towards the rancorous when those boys with their commercial interests get at it
Last edited:
Here are some pointers that came up in my mind, just some unorganized free floating thought.
I wouldn't go over 200K in freq or so, just so you don't have to use RF braid for the heater wires. But i'd have to look up the skin effect dept for 18-20AWG OCF.
I think if your going to heat the tube, you want to use a high frequency tranformer with a center tapped secondary, however that also means that the DC component will flow through your HF transformer, and this could mean trouble.
Another way is to use a half bridge with a LC on the switching node output.
LT sells the LTC6962 which i used for my high school project, you could output that into a HIP2100 to do a half bridge. Do a LC filter plus some 1206 ferrite, cause your not building a FM heater.
The LTC6992 can be modulated by a run of the mill LM393 comparator, I'd just put a pot +c over it to set the time constant of the feedback loop.
From what i remember from class, You'd be looking at a differential equation with two time constants, if you choose the analog part to have the high time constant and everything.
There are some good fast TrueRMS converters that could be used for feedback purposes, but the the flipside is that this adds a little bit of phase shift.
I wouldn't go over 200K in freq or so, just so you don't have to use RF braid for the heater wires. But i'd have to look up the skin effect dept for 18-20AWG OCF.
I think if your going to heat the tube, you want to use a high frequency tranformer with a center tapped secondary, however that also means that the DC component will flow through your HF transformer, and this could mean trouble.
Another way is to use a half bridge with a LC on the switching node output.
LT sells the LTC6962 which i used for my high school project, you could output that into a HIP2100 to do a half bridge. Do a LC filter plus some 1206 ferrite, cause your not building a FM heater.
The LTC6992 can be modulated by a run of the mill LM393 comparator, I'd just put a pot +c over it to set the time constant of the feedback loop.
From what i remember from class, You'd be looking at a differential equation with two time constants, if you choose the analog part to have the high time constant and everything.
There are some good fast TrueRMS converters that could be used for feedback purposes, but the the flipside is that this adds a little bit of phase shift.
VL, what temperature shock? I think both you and Rod are muddleheading us, recrystallization and now temperature shock. GM70 works fine with plain AC-powered filaments and lasts thousands of on/off cycles. This is how it was intended to be used by the manufacturer.
You are absolutely right in your assertion that AC is perfectly adequate.
However i have had some Philips/Mullard ECC here that have the characteristic flash during startup, and I have had tubes that weren't gassy where the filament flashed and burned up immediately.
I think but i dont know for certain, so take it with a grain of salt if you will. that depends somewhat on how the transformer feeding the tube is dimensioned, A small 50VA EI core feeding a single 10V 3A 845 is gonna dump less current into a cold filament than a 5A winding on a 600VA mains transformer .
From what i know about tube construction, and examining a handful of different types is that the thoriated tungsten fillament either break at the weld, or at the bend where it is held up by a spring at the top.
Some tubes have nickel or nickel-iron on tungsten spotwelds, and those two metals have different expansion characteristics. So that could explain weld failures.
I have to finish the last set of modules designed for 845, If my chaotic *** remembers to do so, i will try to make a video of how the tube starts. Its like you use a dimmer on some halogen bulbs indoors and its quite nice and a noticeable difference.
PS: You wont hear me about pervasive mains induced noise, or the likes. Plus i have designs online for free. for 2.5A max and 15A max. Yes i sold some of those at a profit, to finance my hobby.
Edit, ever notice how a lightbulb burns out when you switch it on and not during use? that is pure tungsten... I remember seeing a video of a Firehouse that had a single bulb running for over 100 years. This is purely anecdotal...
However i have had some Philips/Mullard ECC here that have the characteristic flash during startup, and I have had tubes that weren't gassy where the filament flashed and burned up immediately.
I think but i dont know for certain, so take it with a grain of salt if you will. that depends somewhat on how the transformer feeding the tube is dimensioned, A small 50VA EI core feeding a single 10V 3A 845 is gonna dump less current into a cold filament than a 5A winding on a 600VA mains transformer .
From what i know about tube construction, and examining a handful of different types is that the thoriated tungsten fillament either break at the weld, or at the bend where it is held up by a spring at the top.
Some tubes have nickel or nickel-iron on tungsten spotwelds, and those two metals have different expansion characteristics. So that could explain weld failures.
I have to finish the last set of modules designed for 845, If my chaotic *** remembers to do so, i will try to make a video of how the tube starts. Its like you use a dimmer on some halogen bulbs indoors and its quite nice and a noticeable difference.
PS: You wont hear me about pervasive mains induced noise, or the likes. Plus i have designs online for free. for 2.5A max and 15A max. Yes i sold some of those at a profit, to finance my hobby.
Edit, ever notice how a lightbulb burns out when you switch it on and not during use? that is pure tungsten... I remember seeing a video of a Firehouse that had a single bulb running for over 100 years. This is purely anecdotal...
Last edited: