Boost Regulator Modules and Circuits.

[Ferric Toes]

MT 3608 input and output capacitors

The MT3608 data sheet states that input and output capacitors of 22uF are recommended. I thought I would check to see if that was so on the modules I have purchased. I originally purchased 4 off and then purchase another 4 this week from the same supplier.
It appears that the second 4 maybe have a different manufacturer in view of the printing on the underside. Otherwise the topside layout is the same. There is a considerable difference in the capacitance values between the two purchases as can be seen in the pictures, with the second batch (top of picture) being within the plus/minus 20% tolerance for this type of capacitor. Those in the bottom row are all way outside the recommendation with one that is now boxed up that measured 4.7uF. It is a fiddle to do but I managed to piggyback some 47uF capacitors on the existing ones on a couple of modules, hence the 55uF reading on the capacitance measurement.

Other differences are that the higher capacitance modules have a link on the underside strapping the wiper to one end of the trimmer. Missing track on one of the board layers I wonder? The maker's name on the trimmer is BOCHEN on one and BONENS on the other. A cheap copy of a BOURNES trimmer no doubt, hence the backlash, although the BONENS one seemed better.

To make life easier than piggybacking, I have mounted 10uF through hole ceramic capacitors across the input and output pads in addition to whatever SM values are on the board. Conducted noise levels seem low enough now.

One question regarding my 47uF SM ceramic caps is this graph in the data sheet. What is this bias voltage? On both input and output caps there is obviously a dc voltage, and at first glance it appears that at 15V the capacitance is 90% down. Can't be, surely?


Eric.

[bklein]

.
One question regarding my 47uF SM ceramic caps is this graph in the data sheet. What is this bias voltage? On both input and output caps there is obviously a dc voltage, and at first glance it appears that at 15V the capacitance is 90% down. Can't be, surely?
[ATTACH]37249[/ATTACH]

Eric.

Yes it is true. At a previous employer I studied this with an HP LCR meter that allowed DC bias testing. We resulted in creating a spreadsheet of all ceramic caps and effective capacitance with DC bias. We ended up using a good result 10uF 25V cap - paralleling it where necessary to get the higher capacitance. As we used them all over our pricing went down. 5-10uf much cheaper than 1- 47uF and ESR better. Both TDK and Samsung post DC bias performance on their caps now.

[Qiaozhi]

One question regarding my 47uF SM ceramic caps is this graph in the data sheet. What is this bias voltage? On both input and output caps there is obviously a dc voltage, and at first glance it appears that at 15V the capacitance is 90% down. Can't be, surely?
[ATTACH]37249[/ATTACH]

Eric.

There is a good explanation of the effect of DC bias on high dielectric constant capacitors here ->
http://www.murata.com/en-eu/support/...mlcc/char/0005

[6666]

My MT3608 modules match the lower modules in Eric's picture, the caps measure about 5uF
the general schematic seems to match this picture.

[Ferric Toes]

Yes it is true. At a previous employer I studied this with an HP LCR meter that allowed DC bias testing. We resulted in creating a spreadsheet of all ceramic caps and effective capacitance with DC bias. We ended up using a good result 10uF 25V cap - paralleling it where necessary to get the higher capacitance. As we used them all over our pricing went down. 5-10uf much cheaper than 1- 47uF and ESR better. Both TDK and Samsung post DC bias performance on their caps now.

I tried some dc bias on the 10uF caps that I have just bought. The caps are - Murata Radial RCE 10μF Ceramic Multilayer Capacitor, 50 V dc, +125°C, X7R Dielectric, ±20%, Through Hole.

I don't know how the bias is applied in the HP LCR meter that you used, but I did it via a 1.0Meg resistor from a bench psu. With no bias the meter read 9.3uF. The bias was then increased in 5V steps up to 20V. The capacitance increased at each step up to 10.1uF. To check that the bias was not influencing the meter, I disconnected the capacitor, applied the bias and the meter read zero uF.

I tried the same test with the SM caps on the inverter module where I start at 55uF because of my piggyback 47uF. The capacitance increased again as I repeated the 5V steps. At 20V I had gained 9uF.

These results are the opposite of what has been discussed, but good if true.

Puzzled Eric.

[Qiaozhi]

This is most likely happening because you're effectively trying to measure the value with the capacitor in circuit. The DC bias is causing erroneous results on the meter.

Apparently (haven't tried this though) you can add series capacitor (cap1) of a known value in series with the capacitor (cap2) you want to measure. Then apply the DC bias to cap2 via a 1M resistor, and measure across both capacitors in series. A simple calculation can then be made to determine the value of cap2 from the series measurement. Might be worth a try.

[Skippy]

Yes, that's pretty much what I was going to suggest - put a decent 100uF tantalum bead or low-leakage electrolytic in series with the test cap, see if that gives different results.

[bklein]

I first used a boat anchor HP LCR meter that had a test fixture that allowed DC bias testing. The spreadsheet was done on their later model that also has support for it. Some newer (desktop) meters have the DC bias test incorporated as standard.
Perhaps you could connect a waveform generator or pulse generator with DC offset through a resistor to a test cap and see the effects in the rise/fall times on a scope. It is a factor for multilayer ceramics not electrolytics so try your test on an electrolytic and value should stay relatively the same.
You still have to ask DC/DC converter vendors about their reference designs as some show one cap at 10uf at an output or input and others show 4 of them or more sometimes. I found this effect on my own and gave a presentation to our engineering team. Later the cap vendors owned up to the effects and started supporting it in their documentation. Some power device vendors took awhile to realize the behavior. Here's a good simple report: http://www.niccomp.com/help/VoltageC...-032012-R1.pdf

Barry

[green]

An attempt at measuring capacitance at bias volts. Short test capacitor, open and record charge volts as capacitor charges. Thought the X7R capacitance might change more than it did after searching ceramic capacitors. Simple circuit, maybe not giving the best answer. I have some other capacitors to test unless someone has a different method.

[Ferric Toes]

My MT3608 modules match the lower modules in Eric's picture, the caps measure about 5uF
the general schematic seems to match this picture. [ATTACH]37253[/ATTACH]

Your schematic shows 2.2uF yet the capacitors in the picture are fatter than on the modules I have. The full data sheet for the MT3608 recommends 22uF. Maybe the Chinese just fit whatever comes to hand each day

Eric.

[Altra]

After reading this thread I ordered a few of these modules. The caps on mine measured 20uf and 18uf in circuit

[6666]

After reading this thread I ordered a few of these modules. The caps on mine measured 20uf and 18uf in circuit

There appears to be a variety of the same pcb , looking closely at Altra's pcb his has 3 vias between the input pins and 3 between the output pins were the pcb's that I have there is only 2, maybe whoever made Altra's have used almost the closest cap values.
Altra do you have a link to where you bought yours thanks.

[green]

An attempt at measuring capacitance at bias volts. Short test capacitor, open and record charge volts as capacitor charges. Thought the X7R capacitance might change more than it did after searching ceramic capacitors. Simple circuit, maybe not giving the best answer. I have some other capacitors to test unless someone has a different method.

Tried measuring the 10u X7R using method suggested in replies #36 and #37 using the 100u aluminum capacitor in series, reply #39. Took awhile to stabilize to a 4.7uf reading. Tried the circuit reply #39 with capacitor volts starting at 12v and recording the discharge. Around 5uf. Not confident with either method yet.

[Qiaozhi]

More fuel for the fire ->
http://www.cliftonlaboratories.com/c...age_change.htm

[Altra]

There appears to be a variety of the same pcb , looking closely at Altra's pcb his has 3 vias between the input pins and 3 between the output pins were the pcb's that I have there is only 2, maybe whoever made Altra's have used almost the closest cap values.
Altra do you have a link to where you bought yours thanks.

I ordered mine from here in the US, so it wouldn't take 2 weeks to receive.

http://www.ebay.com/itm/222089094151...%3AMEBIDX%3AIT

Based on a 10 - 20% tolerance and in circuit testing these caps are probably 22uf.

Add:
It would be easy to change the caps

http://www.ebay.com/itm/50pcs-1206-C...oAAOxykmZTOaD6

Also photo of pcb bottom