Simulated load for Vreg

[rzenc]

In order to simulate the load of 12A_7 heaters operating @ 12V/150mA, Can I use a 80 ohms/5W resistor?

Thanks!

Best Regards

R.

[jazbo8]

Yes, you can or if you plan to do more load testing at various currents, try IXCP10M45S on a heatsink with the appropriate current setting resistor.

[rzenc]

Howdy!

I did the simul and went ahead to install a 12AX7 to see how it would behave...so far so good... I'm getting a steady reading of 11.83Vdc between pins 4 & 5. Then I measured how much current was being drawn and got 143.3mA.

According to datasheets I should have 12.6Vdc @ 150mA, which is 1.89W. From my readings I'm getting 1.695W, which translates as 89.68% of specified heater wattage.

Tolerances:

datasheet from Sylvania 12AX7/6AX7 does not say anything besides heater warm-up time with average voltage of 80%.
RCA 6681/12AX7 suggests regulation no more no less than 20% for series heaters running 12V. Mine is 93.89% of specified heater voltage.
Checked other datasheets but found nothing regarding heaters voltage regulation.

I also took some readings with my scope and got about 1mV p-p of noise - pretty awful waveforms with 5mV/div + gain x5 - while probing in parallel with heaters pins.

So my question is: How much ripple/dirt is acceptable when running DC heaters @ 12V?

Thanks in advance!

Best Regards

R.

[sluckey]

1mV pp noise on a 12 volt supply! I'd call that pure DC. Do you have a schematic of your power supply?

[rzenc]

Do you have a schematic of your power supply?

Attached!

1mV pp noise on a 12 volt supply! I'd call that pure DC.

So I guess it's acceptable!

However, I've run into thermal issues when I try to supply more than 1 tube with it. Vreg chip shuts down when temp rises above 123ºC.

I'm running it for nearly 4 hours non-stop. With only one tube it has stabilized on 100ºC without heatsink.

Thanks Steve!

Best Regards

R.

[sluckey]

Put it on a proper heatsink and it should handle up to 1 amp current. That's 5 or 6 12AX7s!

BTW, the LM7812CK is available in a TO-3 case. It can tolerate higher temps and has a higher output current capacity.

[rzenc]

Put it on a proper heatsink and it should handle up to 1 amp current.

Do you know how can I calculate an appropriate heat sink?

BTW, the LM7812CK is available in a TO-3 case. It can tolerate higher temps and has a higher output current capacity.

Did not fin it at mouser, however they suggest NTE933 as cross-reference.

http://www.mouser.com/ds/2/300/nte933-89551.pdf

Thanks Steve!

R.

[sluckey]

I would not attempt to calculate a heatsink. I'd just get one that's about 1.5"x1.5" or 2"x2" with simple fins. Better yet, if the 7812 will be used with an aluminum chassis, just bolt it to the chassis. You must use a mica insulator (or other proper electrical insulator) that's appropriate for the case style of the chip (I assume TO-220 style).

[rzenc]

Al chassis for sure...

I bought some micca insulators, although I heard they will eventually crack.

You are right in assuming TO-220 case.

I was asking about a proper sized heatsink for the big bad NTE933 which could solve many problems regarding placement...

Actually, I was thinking about a full series of 12AX7 tubes (6 tubes in this case), which should gimme 75.6V @ 150mA's. What do you think? Doable?

Best Regards

R.

[sluckey]

I bought some micca insulators, although I heard they will eventually crack.

I would not be concerned. Mica is fragile and if you flex it between your fingers, it will crack or even outright break in two. Once they are sandwiched between the TO-220 and chassis they will be safe forever.

proper sized heatsink for the big bad NTE933

Just mount it to the chassis also. The case is the ground terminal, so no need to use a mica insulator. I'm assuming you are using a separate transformer to feed the 12V supply?

Actually, I was thinking about a full series of 12AX7 tubes (6 tubes in this case), which should gimme 75.6V @ 150mA's. What do you think? Doable?

You're not going to put the tubes in series. They will be in parallel. So, for 6 tubes, you need 12V @ 900mA. That's doable even with the TO-220 chip, but you gotta use a heatsink and grease. That NTE933 would do it easily and would simplify mounting. It might even run cool! But, it costs about $50 and may be impossible to find. I'd just do the TO-220.

[rzenc]

I'm assuming you are using a separate transformer to feed the 12V supply?

Yes, 120/15@3A.

So, for 6 tubes, you need 12V @ 900mA. That's doable even with the TO-220 chip, but you gotta use a heatsink and grease.... I'd just do the TO-220.

I tried 4 tubes (12V @ 600mA) and it did not work out...even with 2" x 2" Al heatsink and grease, TO-220 LM7812A went too hot too fast (123°C in less than 1 minute) and thermal shutdown occurred. 
I ended up with one chip for each tube and it's surprisingly quiet.

You're not going to put the tubes in series. They will be in parallel.

Ok..I believe you might have misunderstood my intention.. I was considering for a moment to run all 6 tubes heaters in a series string with 75VDC @ 150mA instead of 6 parallel 12V @ 900mA heaters. I could simplify things even further... what you think? doable?



Kinda out of subject here but I must ask: Is it reasonable to run power tubes heaters in series? for instance 4 EL34's will eat 24VAC @ 1.6A while the same quartet running parallel goes 6.3 @ 6.4A...

Thanks!
Best Regards Steve!!

R.

[sluckey]

I tried 4 tubes (12V @ 600mA) and it did not work out...even with 2" x 2" Al heatsink and grease, TO-220 LM7812A went too hot too fast (123°C in less than 1 minute) and thermal shutdown occurred. 
I ended up with one chip for each tube and it's surprisingly quiet.

I'm surprised that the chip could not handle a 600mA load. That's definitely not up to advertised specs of 1A. Is it possible you have a defective chip? Did you try more than one chip?

Ok..I believe you might have misunderstood my intention.. I was considering for a moment to run all 6 tubes heaters in a series string with 75VDC @ 150mA instead of 6 parallel 12V @ 900mA heaters. I could simplify things even further... what you think? doable?

Yes, I misunderstood. Series filaments is an option, but I can't think of a single advantage. I personally hate series filaments. That goes back to my tv days of working on series filaments. What a bitch trying to find the oddball open filament while crouching behind a tv set in a customers house! I suppose a simple guitar amp would be easier, but I'm prejudiced now. And don't even get me started on Christmas light strings! 

Is it reasonable to run power tubes heaters in series?

Sure, it's reasonable. But can you show me a single name brand amp that does this?

There is a simple solution to increase the current capacity of your LM7812A. All you need is a 2N3055. Connect the collector to pin 1 (unregulated input) of the reg. chip. Connect the base to pin 3 (regulated output) of the reg. chip. The emitter becomes the regulated 12vdc output but the current capacity will be several amps, probably only limited by your transformer current capacity. You would need a heatsink for the 2N3055. The transistor operates as a series pass (AKA emitter follower) and allows the regulator chip to operate into a lighter load.

[PRR]

> show me a single name brand amp

Is Peavey a brand?

With 0.6++V diodes, getting 6.3VDC means considerable waste. Upping the heaters to 12V or 36V is more efficient. Also working 16V caps at 6V versus 50V caps at 36V, better buy.

Series, there's just one wire from socket to socket, not two.

I hear you about series-string failures, and diagnosis.

Also in a high-feature guitar amp, especially on the bench, there will be times you want to run it one tube out, but in series you can't.

OTOH, in series you can stop futzing with silly-state chips and their heat problems, and regulation for NO good reason at all. (All the incandescent lamps in your house will blow before utility surge insults your tube heaters.) Pick a good-size VAC, rect to DC, cap-resistor-cap filter and dropper.

(Can't do this with parallel because you "can" take a tube out but this will cook the others.)

I hope you are not seriously going to feed good DC to power tubes?? At power tube signal levels, any hum from AC heat is just bad layout.

[rzenc]

Steve:

Is it possible you have a defective chip? Did you try more than one chip?

I tried 4 I think..then I realized my design was poor and I should manage thermal trouble in order to get it working properly....I became suspicious when I tried only 2 tubes (300mA) and it kept overheating faster than we can say to hell with sand!!

I personally hate series filaments.

I understand clearly what you mean and couldn't agree more when PRR says:

"I hear you about series-string failures, and diagnosis. Also in a high-feature guitar amp, especially on the bench, there will be times you want to run it one tube out, but in series you can't."


There is a simple solution to increase the current capacity of your LM7812A. All you need is a 2N3055. Connect the collector to pin 1 (unregulated input) of the reg. chip. Connect the base to pin 3 (regulated output) of the reg. chip. The emitter becomes the regulated 12vdc output but the current capacity will be several amps, probably only limited by your transformer current capacity. You would need a heatsink for the 2N3055. The transistor operates as a series pass (AKA emitter follower) and allows the regulator chip to operate into a lighter load.

Next protoboard project on the way! I will try to simulate a load of 10 preamp tubes which might cover future needs

PRR:

OTOH, in series you can stop futzing with silly-state chips and their heat problems, and regulation for NO good reason at all. (All the incandescent lamps in your house will blow before utility surge insults your tube heaters.) Pick a good-size VAC, rect to DC, cap-resistor-cap filter and dropper.

That was my initial thought about going series heaters @ DCV...but I'm not sure if this will help in preamp since sometimes we need to service amps and unplug some tubes while others are operating. Series string will be upset by the lesser load and probably fry the other heaters. Would need some special care while servicing..maybe too much trouble...

I will sticky to the idea of 12VDC feeding tubes in parallel.

I hope you are not seriously going to feed good DC to power tubes?? At power tube signal levels, any hum from AC heat is just bad layout.

I have never had problems with hum from power tubes. But as you say, one wire to the next socket is far easier than tightly twisting fat wires to handle up to 8A. Can it be done with AC source? Say 25VAC @ xxA?

EL34 bottles equals 6.3*1.6=10.08W required for lighting heaters. 4 tubes 4*10.08=40.32W. If supplying 25VAC for the series string we would need 1.6A which equals 40W. But something tells me that I must have a bigger current capacity right? How big should it be?2A?3A?

Thanks guys!!!

Best Regards

R.

[PRR]

> something tells me that I must have a bigger current capacity

Why?

There is one reason. The cold resistance is about 4 times lower than the running resistance. Fed from _constant_ voltage source, the "1.6A" heater sucks 6 Amps for a split second, tapering toward 1.6A after a second or so.

Transformers easily handle a 4X 1 second overload. They will sag; if 10% sag at normal load maybe 40% sag at tube start. So 3.8V at switch-on, rising fast toward 6V.

Regulators may have a hard current limit. Say you found a 2.0 Amp regulator. At start, the tube might only get 2V 2A. But that's enough to heat it and drive the resistance up.

Resistor is a soft current limit.

Neither regulators nor resistors are attractive for Power tube heating. You don't need low-low hum when tubes need 45V signals. You don't want added loss if you have 40 watts of "necessary loss".

__________________________
Off-topic: the earliest work on vacuum electronics was done with intense UV light on Zinc, not thermionic cathodes.

H. J. van der Bijl, "Verh. d. D. Phys. Ges.," May, 1913, page 338. In these experiments which were also performed under such conditions that the current was carried almost entirely by electrons, the source of electrons was a zinc plate subjected to the action of ultra-violet rays. It is obvious that the action of the auxiliary electrode is independent of the nature of the electron source. Hence the results then found apply also to the present case.

[rzenc]

Ok... I found some interesting readings to do and some more Vreg chips worth checking: LM1085 familien:

http://www.datasheetcatalog.org/datasheet/nationalsemiconductor/LM1085.pdf

Heatsink:

http://sound.westhost.com/heatsinks.htm

I'm still working on a bench protoboard with this idea....found a 1"x 1" Al heatsink on junk parts box and installed it on reg chip (LM7812CV) supplying only 1 preamp tube. Before heatsinking Vreg chip I checked temp on chip surface with thermopar and got a whooping 100°C. After heatsinking it, left it on for almost 12 hours and got a nice 53°C

Next is trying to simulate full load on rectos and check how much PT will sag. 6 preamp tubes should be a load ~13R@10.8W.

Best Regards

R.

[rzenc]

Smoke Test today - literally!

Took reg chip out and simulated a load to measure how much PT would sag. See attached schema for details.

LM1085-12 advises for max input voltage of 18V. Got 17Vdc with 2V_pp ripple.

Room Temperature: 28°C. After 5 minutes resistors went up to 216°C.

17Vdc @ 1.545A suggests a power diss. around 26.3W.

To support 10 preamp tubes (12A_7 family) I will need 12Vdc @ 1.5A, which product gives 18W.

I'm waiting for the chips to arrive in order to start tinkering with heatsink.

Best Regards

R.

[PRR]

You seem to be mixing-up the power in the Load (heaters) with power in the regulator.

*IF* you still had 17V, with 1.5A of heater load, the regulator only feels the 17V-12V= 5V, times 1.5A, is 8 Watts (not 18 Watts).

8 Watts steady is still a hefty load for a TO220 package. It can do that, but you may not have a suitable heatsink. I would be shopping for 20-Watt power amplifier carcases,

[rzenc]

You seem to be mixing-up the power in the Load (heaters) with power in the regulator.

Got it!

LM1085-12 arrived today! 

I can input 15Vdc into it...which might took away around 3W out of reg chip...

So, 18V-15V=3V. 3V/1.5A = 2 ohms @ 10W (this R must be adjusted to suit specific circuit under test).

Now chip feels 15V-11.76V (worst case according to datasheet) = 3.24V. With 1.5A of load, total diss in chip should be 4.86W.

I will try it. I will also try to find a cooler which may fit heatsink. Will report back!

Thanks guys!

Best Regards

R.

[rzenc]

I was studying LM1085 datasheet and find something interesting....according to datasheet the tab which we screw heatsink also appears to be a v out??

If so, heatsink must be isolated from ground/connections contact otherwise we might get fireworks??

[sluckey]

You don't have to insulate the whole heatsink. Just insulate the chip from the heatsink with a TO220 mica insulator and nylon screw/nut. If nylon screw is not available, you can use a metal screw with a shoulder washer.

[rzenc]

Today I tried another set up and so far it seems to be running good....

See attached schem for details.

Regarding temperature measurements... where exactly should I measure temperature in the chip? I ask this because I'm having some difference between case (65°C) and metal tab (76°C).

I'm employing LM7812 for this circuit. It's supposed to simulate the load presented by 4 preamp tubes running @ 12V/150mA each.

EDIT: forgot to mention, but I'm getting 4mVpp of ripple out of the reg chip.

Thanks in advance!

Best Regards

R.

[rzenc]

UPDATE:

Finally found a larger heatsink (2" x 3") and tried it. I was expecting 5°C to 7°C less... turned out I got a nice 41°C on center of chip and 53°C on metal tab. Temp reduction on center of chip (65°C - 41°C = 24°C) and metal tab (76°C - 23°C). It's on for about 12 hours already. Room Temperature 26°C. So chip temp rising 15°C above room temp...

Next: Try to improve 4mVpp on ripple and hope to get less junkie there too. Plan: parallel smaller film caps before reservoir cap.

Best Regards

R.

[sluckey]

Next: Try to improve 4mVpp on ripple and hope to get less junkie there too.

It's called a battery!   

Congratulations on getting it going.

[rzenc]

It's called a battery!

 

Congratulations on getting it going.

You guys helped me a lot!!!

Now...about the junkie....I installed 3 film caps before reservoir, 100nF - 4n7F - 10pF. While scoping it, I noticed there is a little less junk/noise, but still 4mVpp of gargabe...it's pretty much 120Hz of fundamental plus added noise... Any thoughts on what could be done to further improve it?

Thanks in advance!!

Best Regards

R.

[sluckey]

Any thoughts on what could be done to further improve it?

Seriously, you are done. 4mV_pp ripple is nothing. You probably have more ripple than that with the probe connected directly to chassis ground. Take a look.

[HotBluePlates]

Any thoughts on what could be done to further improve it?

Seriously, you are done. 4mV_pp ripple is nothing. You probably have more ripple than that with the probe connected directly to chassis ground. Take a look.

4mV p-p = 2mV peak = 1.4mV RMS

Hewlett Packard's original AC voltmeter (the 400A of 1942) had a minimum range of 30mV full scale. Your 1.4mV would only be one-and-a-half minor divisions above zero on the meter scale.

It took HP 6 more years to come out with the 400C which could do 1mV full scale. The issue was in part that the electronic amplification needed to register such small a.c. voltages added its own noise, masking the voltage being measured.

So even HP users in the 40's would probably agree with Sluckey that you have essentially clean d.c., given the working voltage of the circuit is so much higher. Your noise/ripple is only about 1/100th of 1% of the total supply voltage.

[rzenc]

Thanks guys!

Now moving to LM1085-12 and a supply to support ten 12A_7 series.

More to come!!!


You guys rule!
Many thanks for your kind attention and support!!!

Best Regards,
R.

[sluckey]

Now moving to LM1085-12 and a supply to support ten 12A_7 series.

Just remember, don't get cheap on the heatsink!

[rzenc]

So I got sidetracked from the mighty LM1085-12 and I'm facing some doubts regarding a new DC supply. This time it will need to be a 6VDC since I will be using 6SJ7, 12AY7 and 12AU7.

3 tubes running @ 6V/300mA each means 6V/900mA. I put my hands on some LM7806 reg chips and tried them on protoboard. See attached schem for details.

My question... Is 50mV of ripple acceptable on such rail?

Best Regards

r.

[sluckey]

yes

[TIMBO]

Hi rzenc, Thanks for doing all the hard work, I'm working on this http://www.el34world.com/Forum/index.php?topic=15795.0 and being a Hi Gain build to split the chassis I think will help isolate the preamp from some of the problems that higains have.

I was wanting to do the preamp heaters in 12vdc and I will have 5 tubes. So your topic is well timed.

I also looked into regulated PSU and have used LM7812 regulators from a local supply but opted for there MC78T12 which is rated @3As. Thanks

[rzenc]

Hi TIMBO!

I checked out MC78T12 datasheet: http://www.datasheetdir.com/MC78T12+download and it seems to have less ripple rejection them LM7812 series. I have never tried these though.

A few observations which might help. 

1) 12A_7 series eat ~150mA when running @ 12V. That implies 750mA. I would use UF5408 (3A) diodes to rectify AC since UF4007 is rated @ 1A. Ample margin for safety reasons;
2) I highly recommend using a big hokkin heat sink (at least 2"x3") with thermal grease. Just a very fine layer of grease on both heat sink and reg chip;
3) To sim load, you will need 16ohm/9W resistor. 2x8ohm/10W (series connected) will do fine, but it will get blistering hot. Watch your fingers mate!;
4) Keep track of chip temperature while you power it up. On my first attemp, reg chip temp skyrocketed seriously fast;
5) I don't know whether you own a scope or not but if so, when probing regulated output voltage from chip, try paralleling more reservoir caps (big 4700uF e-caps) to see how much your filtering will improve. On my last attemp, I had to go from 2x4700uF (200mV pp of ripple) to 7x4700uF to get ripple down to 50mVpp; I also recommend that you do yours on protoboard first, them move it to pcb;
6) You might want to try some small caps (use a factor of 1:100 or 1:1000) to scale your caps. it will improve a lot of high frequency garbage riding along with 120Hz fundamental.

Hope this helps you. And if I can help you further down the road, you are welcome mate!

Best Regards

R.