Boost Regulator Modules and Circuits.

[green]

Some questions. Murata has a graph, capacitance vs volts DC for 50volt caps. If the capacitors were rated 25 volt should I divide the volt scale by 2? Would a 0805 or 1210 size capacitor chart the same graph? For higher value ceramic caps is there a better choice than X7R?

[Ferric Toes]

I have a question too. How do designers cope with the large capacitance variation with bias voltage? If a 22uF is needed, do they put in a larger one to compensate for the capacitance drop due to bias voltage. I have a detector circuit board that at a guess has somewhere between 50 - 100 of these ceramic caps. It must be a nightmare to work out what value caps to put in to compensate for the different bias volts that occur in various parts of the circuit. How do you get consistency in performance between boards?

Eric.

[bklein]

First off many times you don't know for sure if the reference design specifies actual parts or theoretical values unless you ask. If theoretical it is your responsibility to meet it by the right choice of capacitor(s) keeping the DC bias characteristics in mind. I wouldn't think 50-100 caps are all output caps for your board. You typically have both local input caps and output caps on the converters. You can save a bit of money by realizing the input cap to a downstream converter will add to the output capacitance of the upstream one. Best practice is - if you have board space - leave free spots for additional output caps. Downside is when/if one fails. You then need a IR camera to tell which one is shorted. Here is another (large PP) report to confuse everyone more... http://www.kemet.com/Lists/Technical...sSelection.pdf

[Tinkerer]

COG capacitors are a lot better, but cost more.

[ivconic]

Slightly off topic maybe;
but i tried to use MT3608 2A DC-DC Step Up to supply PI detector and burned 4 of those in no time.
And that particular PI detector draws not more than 150mA.
I used it to boost 12V to 18V which is quite alright and in range of specs:

• 2A maximum output current
• 2.0 to 24VDC input voltage range
• 28V maximum output voltage (trimpot adjustable)
• 93% peak efficiency (~200mA output current at 5Vin, 12Vout)

I am confused?
XL6009 based step up did the job fine later. Just fine!

One simple layman question; there are two elco caps on XL6009 based board and there are no elcos at all on MT3608 board !?
Could i save MT3608 from burning if i added elcos ?
I guess i could smooth up the "sudden" drain transients (critical moment at switching ON the detector) if i added elcos?

[Ferric Toes]

This is very much 'on topic'. The MT3608 is exactly the module that I have been trying. I too have burned a couple, but it was my own silly fault. One was when I put a 2200uF across the output when it was switched on, and the other was that I assumed the metal retainer on the choke was at ground potential and ended up by shorting the choke. By the way, your picture shows a broken outer ring on the ferrite.

Aside from the above I have had no failures and always test with an 82 ohm 5W resistor at 15V and 10V input. I also scope across this resistor for noise. Adding a 22uF tantalum bead on the output make no difference to the noise. However, 10uF through hole ceramic gives a big reduction. The lowest noise is with a 68uH 0.3 ohm choke and another 10uF ceramic forming a pi filter.

I must get a XL6009 to try for comparison. I went to the higher frequency MT3608 because I thought that there was less chance of interference with a PI detector. Is your detector PI? Luckily these modules are quite cheap, so if you fry one it doesn't burn a hole in your wallet.

Eric.

[Ferric Toes]

http://www.kemet.com/Lists/Technical...sSelection.pdf

Thank you for this link it was very informative.

Eric.

[daverave]

Hi ivconic...ive used step up converters in two of my pulse detectors with no problems....not sure why your voltage converters are burning...mine are cheap one's which i purchased from ebay.

[daverave]

This is very much 'on topic'. The MT3608 is exactly the module that I have been trying. I too have burned a couple, but it was my own silly fault. One was when I put a 2200uF across the output when it was switched on, and the other was that I assumed the metal retainer on the choke was at ground potential and ended up by shorting the choke. By the way, your picture shows a broken outer ring on the ferrite.

Aside from the above I have had no failures and always test with an 82 ohm 5W resistor at 15V and 10V input. I also scope across this resistor for noise. Adding a 22uF tantalum bead on the output make no difference to the noise. However, 10uF through hole ceramic gives a big reduction. The lowest noise is with a 68uH 0.3 ohm choke and another 10uF ceramic forming a pi filter.

I must get a XL6009 to try for comparison. I went to the higher frequency MT3608 because I thought that there was less chance of interference with a PI detector. Is your detector PI? Luckily these modules are quite cheap, so if you fry one it doesn't burn a hole in your wallet.

Eric.

Hi Eric....in two of my pulse detectors ive used these type of voltage converters and because of no space in my control box ive used a very large capacitor of 4700 uF in my separate battery box across the output of the voltage converter and ive had no converter burn and they work perfect and no noise present from the converter to the detector...i screened the converter module in copper shield.

[ivconic]

This is very much 'on topic'. The MT3608 is exactly the module that I have been trying. I too have burned a couple, but it was my own silly fault. One was when I put a 2200uF across the output when it was switched on, and the other was that I assumed the metal retainer on the choke was at ground potential and ended up by shorting the choke. By the way, your picture shows a broken outer ring on the ferrite.

Aside from the above I have had no failures and always test with an 82 ohm 5W resistor at 15V and 10V input. I also scope across this resistor for noise. Adding a 22uF tantalum bead on the output make no difference to the noise. However, 10uF through hole ceramic gives a big reduction. The lowest noise is with a 68uH 0.3 ohm choke and another 10uF ceramic forming a pi filter.

I must get a XL6009 to try for comparison. I went to the higher frequency MT3608 because I thought that there was less chance of interference with a PI detector. Is your detector PI? Luckily these modules are quite cheap, so if you fry one it doesn't burn a hole in your wallet.

Eric.

Yes, it is (again) FelezJoo PI detector. It consumes between 150 and 165mA at most. I haven't noticed any problem in use XL6009-based step up with it, for longer period of time. No interference... at least not noticeable in common use of detector.
But also, as i said; i burned several MT3608's with same setup!? Referring the specs it should not happen. Most critical moment is moment of switching ON the detector, and than MT3608 breaks for some reason while XL6009 survives.
So i was wondering why this is happening? Maybe there is something that interferes with MT3608 and not with XL6009 ? Or detector is pulling too much current in peak when initially switching ON and MT3608 simply can not handle it?
I burned all 4 pcs i had at the time, hadn't time to check them with additional caps on the input and output.

[ivconic]

Hi ivconic...ive used step up converters in two of my pulse detectors with no problems....not sure why your voltage converters are burning...mine are cheap one's which i purchased from ebay.

XL6009 is also cheap. So i used it because it do the job just fine.
There are two versions of those; "luxury" one and the "small" one.
I am having both. The small one i use with detector, it occupies less space in battery compartment and the "luxury" one as just another bench supply for various experiments.
"Luxury" one is 7.5 euros and "small" one is 2.7 euros here at local suppliers.




In action, but here supplying detector only with 12 volts, as initial experiment with Arduino:

[daverave]

the step up module you show in your photo is the same type i use with no problems.

[Ferric Toes]

I now have a couple of the XL6009 400kHz modules to experiment with. I set one up with an 8V input from a bench psu and an 82 ohm wirewound 5W resistor as load. The output voltage was set with the trimmer to 15V. A scope was connected across the load resistor and the following waveform observed.

This shows the ripple on the output of 250mV pk -pk, which I would consider excessive for my application and would likely cause problems. The main capacitor across the output is 220uF and there is a smaller ceramic for HF bypassing near the converter IC. The addition of a 10uF ceramic cap across the output terminals makes a large difference.

Adding a 68uH choke and another 10uF ceramic to make a Pi network brings the conducted noise right down. Note the change in vertical sensitivity to 10mV.

The 68uH choke may not be optimum but it is only 0.3 ohms resistance and I happened to have a few on hand.

Even with their bias problem ceramic capacitors seem to be the most effective in killing the noise.

Eric.

[green]

I now have a couple of the XL6009 400kHz modules to experiment with. I set one up with an 8V input from a bench psu and an 82 ohm wirewound 5W resistor as load. The output voltage was set with the trimmer to 15V. A scope was connected across the load resistor and the following waveform observed.
[ATTACH]37434[/ATTACH]
This shows the ripple on the output of 250mV pk -pk, which I would consider excessive for my application and would likely cause problems. The main capacitor across the output is 220uF and there is a smaller ceramic for HF bypassing near the converter IC. The addition of a 10uF ceramic cap across the output terminals makes a large difference.
[ATTACH]37435[/ATTACH]
Adding a 68uH choke and another 10uF ceramic to make a Pi network brings the conducted noise right down. Note the change in vertical sensitivity to 10mV.
[ATTACH]37436[/ATTACH]
The 68uH choke may not be optimum but it is only 0.3 ohms resistance and I happened to have a few on hand.

Even with their bias problem ceramic capacitors seem to be the most effective in killing the noise.

Eric.

If you purchased more, which one would you order? MT3608 or XL6009

[Old cart]

I now have a couple of the XL6009 400kHz modules to experiment with. I set one up with an 8V input from a bench psu and an 82 ohm wirewound 5W resistor as load. The output voltage was set with the trimmer to 15V. A scope was connected across the load resistor and the following waveform observed.
[ATTACH]37434[/ATTACH]
This shows the ripple on the output of 250mV pk -pk, which I would consider excessive for my application and would likely cause problems. The main capacitor across the output is 220uF and there is a smaller ceramic for HF bypassing near the converter IC. The addition of a 10uF ceramic cap across the output terminals makes a large difference.
[ATTACH]37435[/ATTACH]
Adding a 68uH choke and another 10uF ceramic to make a Pi network brings the conducted noise right down. Note the change in vertical sensitivity to 10mV.
[ATTACH]37436[/ATTACH]
The 68uH choke may not be optimum but it is only 0.3 ohms resistance and I happened to have a few on hand.

Even with their bias problem ceramic capacitors seem to be the most effective in killing the noise.

Eric.

Are you talking about this being the coil supply then deriving +/- voltages from this , maybe via a ground splitter arrangement? If so you could boost this output to 18 volts or so, regulate to 15 volts via a linear regulator which should give low noise, then rail split. This would work well if you were using separate TX/ RX coils so that the high current coil return would not go through the rail splitter.