We have been designing a very high precision control circuit for quite a while. And we are thinking to make it much easier to use, which is to get rid of the Lab DC Sources and use AC to DC switching power (get power from 220V/110V directly).

Our circuit runs at +-12V currently with around 200mA current. It consists of a photodiode, which 1pA precision could be measured, many OpAmps, one 16-bit ADC and one 18 bit DAC. As from the description, it can be observed that our circuit is running at very high precision.

One lab DC source costs thousands of dollars and the DC ripples are basically zero. At most on the level of 0.001mV peak to peak. If a cheap AC to DC source is to be used, however, I am afraid that the DC can be far less clean. For example this one (which has 50mV pp ripples).

I need suggestions from you:

  1. Is this kind of AC-DC switching power a good choice (http://www.switchingpowersupply.cc/pdf/D-30W.pdf)? Or maybe the ones used in laptops are better?

  2. How to reject the noises in power supply in the most thorough way? Any material will be appreciated.

  3. Is it possible to make clean DC source from 220V AC, which has similar performance as the lab DC sources?

  4. If large caps and inductance are used to filter the ripples, could they cause stability issues for the AC-DC converters?

  5. How would the DC noises influence the behavior of OpAmp, ADC, DAC? For ADC and DAC, I think the voltage references will not be influenced.

  • 1
    \$\begingroup\$ What current needs to be supplied? \$\endgroup\$ – Andy aka Mar 13 '14 at 13:51
  • 1
    \$\begingroup\$ You need a better spec. ("high precision" isn't a spec.) How much ripple is allowed? What is the ripple rejection of the load circuits, both at DC and at higher frequency? \$\endgroup\$ – markrages Mar 13 '14 at 17:13
  • 1
    \$\begingroup\$ There are some good appnotes / papers in the Linear Technology archive about very high precision test, measurement, design, etc. that may be of interest to you. \$\endgroup\$ – John U Mar 13 '14 at 18:00
  • \$\begingroup\$ @markrages You are right. I am not sure about the worst ripple allowed. The only spec I am aware is that the first stage preAmp turning photodiode current into voltage should have the precision of 1pA. To connect ripple allowed and my spec, is "ripple rejection ratio" the term to study? \$\endgroup\$ – richieqianle Mar 14 '14 at 4:43
  • \$\begingroup\$ @richieqianle, yes, often abbreviated in datasheets to "PSRR" for power-supply rejection ratio. \$\endgroup\$ – markrages Mar 14 '14 at 5:55

Usually if you need clean power, you need to follow up the SMPS with a linear regulator. You also have to deal with the noise created by the supply, which can be non-trivial. If you really need microvolt noise levels, a completely linear supply might be a good choice rather than SMPS, (or even use SLA batteries and an ultra-low-noise regulator).

On the other hand, getting 1mV to 10mV p-p noise over a 0.1Hz to 1MHz bandwidth is not that difficult with a SMPS + LDO combination, but care has to be taken in how the SMPS conducted and capacitively coupled noise is dealt with.

Do not underestimate the noise the supply itself can create- I found that the brick supply for a Tek scope can cause 20mV false spikes in the display if the plastic case is placed near a grounded metal object. That's the supply that the oscilloscope comes with! There are transistors (and possibly entire heat sinks) flapping around by hundreds of volts and switching fast at tens or hundreds of kHz.

  • \$\begingroup\$ Thanks a lot! For the last comment of noises created by supply, I am not quite clear about it. Could you explain a little more? As I need to ensure the power solution works for both Asia and Europe, I may choose SMPS instead of completely linear power. \$\endgroup\$ – richieqianle Mar 14 '14 at 5:20
  • 1
    \$\begingroup\$ Here is a brief overview: ridleyengineering.com/… \$\endgroup\$ – Spehro Pefhany Mar 15 '14 at 2:20
  • \$\begingroup\$ SMPS generally add some common-mode noise to the power supply as well, so that your circuit can become sensitive to other things connected to the outside world. Those noise currents are usually imperfectly conducted to the supply through a 'Y' capacitor and/or some earthing. All of this can influence your noise floor in the powered circuitry. \$\endgroup\$ – Spehro Pefhany Mar 15 '14 at 2:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.