I tested the circuit today with a linear supply, and then everything is as I want it; only 2 counts noise in a steady signal in half an hour, with the usual 1 count uncertainty. I run a simple continous sampling test, and keep track of min and max values. With a button I can reset the min/max values to the current value when I change the level on the input.
That's a classic solution, but it's not practical in my application, which is an unusual high-voltage STM preamp that needs +/-120V and two to three other lower voltages. I fear that 240V of rechargable batteries is a bit much!
Some good solutions were put forth in the thread referred to above. Perhaps sometime I'll get back to that project. In the meantime I'm using a small 1W dc-dc converter that creates common-mode noise at 80kHz. Anybody ever come across a three-winding common-mode choke?
I know this may sound over-the-top, BUT if your circuit has the space, you might want to consider doing away with the "noisy" DC/DC converters and use a few NICAD or NIHM rechargeable batteries as the power sources for the relevant voltages concerned. Higher current capacity batteries would do for the higher current areas of your circuit with the added advantage of NO NOISE or extra low noise from the power supply sections. I would design the circuitry such that the batteries are 'recharged' ONLY when the system is not being "actively" used - thereby transferring any noise generating components to the system's IDLE or STANDBY periods.
I read in sci.electronics.design that Winfield Hill wrote (in ) about 'Low noise DC/DC converter', on Fri, 7 Jan 2005:
Not off the shelf, but it can't be that difficult to wind one. A three- winding differential mode choke is quite another matter.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
I need +/- 15V, that's 25 batteries, a bit bulky ;) And I need the isolation.
I have prototyped an isolated supply with the LT3439; TSSOP16 is really nice to breadboard... It has an exposed backside, so I put it on the board on its top; a wire soldered onto the exposed pad holds the chip firmly in place.
No, I only have small inductors at the dc-dc input. Yesterday I wound a three-winding CM output choke, and it helped (about a 4x improvement), but I'm still seeking further improvement.
Despite the output CM choke, the problem continues to be 50ns 80kHz repetition-rate charge spikes into the STM tip-bias, at point x, creating a small but highly-damaging 80kHz sine wave.
I tried moving some of the 3uF HV-filtering capacitance to the dc-dc common, point y, but the result was slightly worse. Hmm, looking at the drawing above, I could add a resistor at point z, and retain most of the filtering capacitance at point x.
Yesterday I sent the preamp on its way to Hong Kong where it'll be placed into an experiment, so that idea and others I'll have to try on the next prototype.
Oops, must not have caught up with your first posts to the NG. YEP...:) 240V of rechargable batteries is DEFINITELY a bit much - sorry for that. Maybe you should have specified what the 'unregulated' o/p voltages (with expected currents) of the DC/DC converter were to be - then we could have provided a more useable solution - what. I suggest you might want to look at the sort of converter circuits used for microwave RF power supplies - probably refer to the latest ARRL Handbook for ideas.
I would be interested in seeing what the general configuration is inside the "noisy DC-DC" box. Especially, is there any capacitors between the common or supply rail on the input side to any of the 3 output nodes (to shunt any inductive parallel capacitance).
In that case, I would have experimented with capacitors directly between input pins and output pins (inside any external filter inductors. An extra output to output common and input to input common set of high frequency capacitors is probably also in order. Since they are trying to minimize the converter volume, I would expect them to skimp on the bypass capacitors.
Yes, I tried that, or more accurately a porti>> Despite the output CM choke, the problem continues to be 50ns
The normal-mode noise on the +/-15V outputs is a single ring of pulse about 150mV in amplitude lasting about 200ns. Although I know well about normal-mode noise, I hadn't earlier thought to set up and measure its energy. Actually, it's charge I need to measure. I did calculate how much current it must be to create the observed 250uV 80kz sine wave on the 3uF at point x (actually calculated from the observed STM preamp output), and the number, 380uA, seems far too high. BTW, I'm well aware of the necessity for careful probing techniques, and have double-checked this. I've also determined that radiated RFI from the switching module is not the problem. I'll be going back to the bench on this one.
Yes, I tried that, or more accurately a porti>> Despite the output CM choke, the problem continues to be 50ns
The normal-mode noise on the +/-15V outputs is a single ring of pulse about 150mV in amplitude lasting about 200ns. Although I know well about common-mode noise, I hadn't earlier thought to set up and measure its energy. Actually, it's charge I need to measure. I did calculate how much current it must be to create the observed 250uV 80kz sine wave on the 3uF at point x (actually calculated from the observed STM preamp output), and the number, 380uA, seems far too high. BTW, I'm well aware of the necessity for careful probing techniques, and have double-checked this. I've also determined that radiated RFI from the switching module is not the problem. I'll be going back to the bench on this one.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.