Charge pump Issue

[maiko]

I built a charge pump to use for a firefly tube amp. tc7660 dioded 1n4148 and 10uf 160v caps

problem is even with all this cascading



Im only getting 25 v (dc) and worst it keeps dropping at a rate of .01v almost per second

Ive tried running the 9v off my 3 amp Powersupply regulated with a lm350.   so its not a supply issue.

is the output suppose to be DC or AC when i switch my DMM to ac i see 50v++ but still its dropping

Any help would really be appreciated I cant really move on to finishing my Firefly without sorting this out first.

[R.G.]

There ain't no such thing as a free lunch.

Pumping charge through a voltage multiplier involves an increase in impedance as well as voltage. A TC1044 has an internal equivalent impedance of about 40-80 ohms as a doubler. Using it as a tripler, quadrupler, quintupler, and more also multiplies that impedance until it's bquite large and any loading makes the voltage sag.

You can help it somewhat by pushing the input HARD by driving it with high current semiconductor switches, driving it very high frequencies, to minimize the impedance of the capacitances involved, making the capacitances larger, and using very low-impedance diodes. It doesn't hurt to run your charge pump as several parallel pumps in non-overlapping phases to contribute more charge pumped.

As you're guessing by now, the things needed to make it work with lower impedance, you need to redesign it from the ground up to be something different from the TC1044 and a few diodes and caps.

Mother Nature is quite demanding about everything following Her laws.

[Phoenix]

1N4148 is also a bad choice for the rectifiers here. Their forward voltage at the surge currents required to charge the capacitors is quite a bit higher than that of a dedicated rectifier diode. At 100mA, their forward voltage is already 1.1V. Even the lowly 1N4001 is still below 0.8V at that current. Schottky diodes like 1N5817 would be a better choice again.

[maiko]

Thank you for the responses will look into the diodes tonight.   would a 5819 or 17 be alright its only 20 - 40 volts respectbly

[Phoenix]

Thank you for the responses will look into the diodes tonight.   would a 5819 or 17 be alright its only 20 - 40 volts respectbly

The diodes only see the voltage difference between the stages of capacitors - 12V (and less at the subsequent stages when accounting for the forward voltage losses), so yes, they'll be fine.

[amptramp]

Charge pumps have their limitations.  They also get quite a large parts count with more levels of multiplication.  For the levels of current you want, a flyback converter will give you ample voltage with much better efficiency since there is only one rectifier drop and it will be a lot smaller.  Other circuit topologies such as push-pull and resonant filters also have their good points but a flyback converter is usable up to at least 50 watts and you are nowhere near that.  Old computer power supplies can be modified to give you tube-level voltages if you are willing to rewind an inductor.

[Ripthorn]

I've done quite a bit with charge pumping in the past and if you are trying to get above 30-40V from a 12V input, I would go with something else.  Even the nixie style power supply works well inside its limits.  Parts for it are relatively cheap.

[Kevin Mitchell]

Quick question;
Why wouldn't you use the Superfly's SMPS power supply?

[thomasha]

I fact the superfly SMPS is better IMHO,

more voltage, more current... have I forgot something?

I like the Max1771, has less components and a good guide here>
http://www.desmith.net/NMdS/Electronics/NixiePSU.html

and the layout is easy, I did a lot of different ones, and they always worked, sometimes better, sometimes not, I could put a list of amps and layouts to compare, but in the end the main problem is always current and heat.

My best version runs a PCL84 at 235V and this one is very stable.

The superfly and the parallel SE version of the same amp work great, but I screwed up making the first stage too close on the superfly mainboard (but it fits a 1590a). The parallel SE is actually a 555 version with 12 mA current measured at the cathode resistor.

The four and 3 tubes version are really nice, but I built it in a small combo, and after 1h playing there is quite a voltage drop. The thing is I need to place the amp with the speaker to the bottom to reduce the heat and voltage drop.

The head version works better! the SMPS is the green board>


If you're interested I can help you with this, and with access to mouser or something similar you can buy good components (with low ESR, without the seller asking you what ESR means...)

cheers,
Thomas

[maiko]

Thank you folks for the reply

But as much as i want to give SMPS a try  i dont have access to the ic MAX1771ESA in fact the max1044 is not available here all i have is the 7660


Project Update:   Tried 5819 diodes it was going well and voltage seemed to be stable but suddenly one of the Caps blew up(as is exploded)  luckily the board was upside down as i had the pcb underside facing be whilst i was taking voltage readings.   I dont understand why it blew.  It was not at the end of the cascade and the caps were rated at 160v   i checked polarity was not wrong.    and end voltage was 30v so there was no reason for it the blow up.

[Phoenix]

Thank you folks for the reply

But as much as i want to give SMPS a try  i dont have access to the ic MAX1771ESA in fact the max1044 is not available here all i have is the 7660


Project Update:   Tried 5819 diodes it was going well and voltage seemed to be stable but suddenly one of the Caps blew up(as is exploded)  luckily the board was upside down as i had the pcb underside facing be whilst i was taking voltage readings.   I dont understand why it blew.  It was not at the end of the cascade and the caps were rated at 160v   i checked polarity was not wrong.    and end voltage was 30v so there was no reason for it the blow up.

There was a reason for the blow up... Ripple current.
You'll need to use low ESR capacitors if you want to proceed down this path, which you're already discovering probably isn't the best method.

[maiko]

Going back to the charge pump

according to this c1 should be (-) to pin2

http://www.diystompboxes.com/smfforum/index.php?topic=75457.0

But looking at the 1044 pinout pin 2 is cap+   but then looking at the way he solved the problem and pic it looks at though he did fliped the cap and it solved the problem.   

Im having exactly the same problem as he is

[Phoenix]

Going back to the charge pump

according to this c1 should be (-) to pin2

http://www.diystompboxes.com/smfforum/index.php?topic=75457.0

But looking at the 1044 pinout pin 2 is cap+   but then looking at the way he solved the problem and pic it looks at though he did fliped the cap and it solved the problem.   

Im having exactly the same problem as he is

Check the datasheet!

For a voltage inverter, pin 2 should have the caps positive terminal attached, but for a voltage multiplier, pin 2 should connect to the negative terminal of the cap.
If you had a cap in backwards, it's no surprise it blew.

[maiko]

WHEEEEEEE

It works now Flipped em caps and not it works.   4 cascades and its giving me a solid 45v   6 should easily get 75 

Thanks to all the people who have given there inputs to this.    Im grateful .   This is just the first steps though now to build the amp. 

will keep you guys posted.       

Still interested in the nixie though and will experiment more on  that.

Thank you to all

[thomasha]

I guess you have read it already, but the 7660 has a 7660S version that works better.

if you have access to mouser you could buy low ESR caps, low voltage drop ultra fast diodes that would increase your voltage a little.

The problem with charge pumps is the 10 mA max. current limitation (at least the max1044 has it), 10 mA is to low, when compared with the results you obtain from a SMPS. Even the 555 can deliver more. I'm not sure how much a firefly will need, but I have PP subminiature amps that are biased hotter.

But, I saw some really nice amps with it too, with 1/4W output, so please continue and post your results so we can have an idea of how it's working.

[amptramp]

For those of us who don't like charge pumps for small tube amps: