Parts issue, Benboost and more...

When putting together a parts list for the Benboost, I found what I thought was the holy grail of 0805 capacitors. It was a Kemet x7r, 22uF, 16v 0805 capacitor from Allied Electronics.

It was on sale and seemed to be the perfect capacitor for output on many drivers, especially the BenBoost.

In fact, at one point there was some question about whether Benboots were killing 12x 405's and single mode blue diodes.

So, I took a Moh version of the Benboost, with 10uF capacitors on it and tested it on a scope, then wiped them and installed the Kemet 22uF caps and saw a great improvement.

Meanwhile, LeQuack built a 405 using parts for the Benboost including the 22uF Kemet capacitors from Allied Electronics.

Recently Foumist (Tom) asked to buy some driver parts from me. I threw some of these caps in with the package as a gift.

Tom wrote me today to earn me that he tested them and they were only 2.2uF? I doubted this, since I had ordered them from a reputable supplier, but I tested them myself and got 2.2uF as well.

I contacted LeQuack and he did some research and the short version is that Allied missed by a decimal point. They did this on TWO versions of the capacitor. A 20% and 10% version were both listed as 22uF, 16v x7r caps. In reality, they are 2.2uF, 16v, x7r caps. All of them, even though they are two different part numbers and two different tolerances. Allied screwed up... And I did not think to check them...

Now, here is where it REALLY gets weird.

First, LeQuack has been using these without problem in his 405nm 12x build.

Second, I recently tested the Benboosts and they did better with the 2.2uF caps than with the 10uF caps on output.

Third, LeQuack tested them on a different driver as well. They performed BETTER than the 10uF caps on this driver as well.

He did some asking on other sites and it has been suggested that since we are designing constant current drivers based off of constant voltage chips and schematics, maybe the higher capacitance is slowing the ripple frequency (which would explain the slower than expected frequency on the benboost) and creating more heat.

I wanted to put this out there, I don't know who I have sent all these parts to, but if you have caps from me, they may be incorrect,

Also, they may work! We are trying to figure out why this is happening.

I am wide open to discussions...

Not really slowing the ripple frequency, but causing oscillations, and maybe a little bit of resonance which would REALLY cause things to screw up and perhaps kill diodes. And the reason I say oscillations and resonance is because if you take a gander at the ripple output of the drivers, it's a sine wave... Sine waves are caused by resonance + ringing which can be caused by stray oscillations caused by the driver freaking out over the excessive output capacitance. While working on a prototype of one of my drivers I noticed when I was using 2.2 uF capacitors, the output ripple was basically non-existent. Switching over to 10 uF capacitors, I noticed the ripple took a huge rise and was a whopping 1 volt peak to peak sine wave... And the driver IC I was using was getting REALLY hot while outputting 1 amp.

Using only 2.2 uF caps, the chip barely even got warm.

It's definitely something to look into, and I trust the guys over at 4hv, and they've given me plenty of tips and pointers to help me out when designing my new laser driver.

Hmm. Interesting. May have to investigate this on new drivers. I've been assuming more capacitance is better because all beliefs would lead to that conclusion... yet I haven't actually compared output capacitance vs. ripple for the BlitzBuck. For this reason, I have been working on increasing capacitance on the BlitzBuck, but it may not help! I'll send 'em over to you, Tom, to test!

Weird. I just got a scope so I can do my own testing when I get back home

I hope this isn't entirely true, got new driver PCBs coming in made for 200uf of output capacitance!

Lower caps than recommended will not be that good with other ICs, though.

If you look at the LM3410 Datasheet the output is recommended to be at least 0.47uF. There is a function for calculating the desired capacitance at certain voltage and inductor current. Most of it's application in the datasheet use 2.2uF or 10uF output caps.

Overthrowing the needed capacitance might result in a increased spike on startup BUT choosing a lower value than needed will result in a spike also but lower one. If you look at the ben boost thread where Tom (tsteele93) posted scope pictures you will see there is a spike with both 10uF and 2,2uF caps. But the 2,2uF caps the spike is lower than the overall output current so it might be considered SAFE. But it's still there. See here:

10uF caps:

2,2uF caps:


see both have spikes. AND the ripple is still great. It might look lower but it's still not THAT better.

I suggest you do the following 3 test:

* Use a 22uF cap between Vout and Ground. And remove the cap between Vout and FB (LD+ and LD-).
* Use a 10uF cap between Vout and Ground. And remove the cap between Vout and FB (LD+ and LD-).
* Use a 2,2uF cap between Vout and Ground. And remove the cap between Vout and FB (LD+ and LD-).

compare the results (Be sure to remove the Vout / FBpin cap!)!

I do believe you will get a better output compared to both 10uF caps and both 2,2uF caps.

Ok, adding too much capacitance on the output will result in current spike on startup along with smoothing the current ripple.

To fix the spike you must also add an appropriate cap value between Vout and FB pin. That will create a slower ramp up of the output current, smoothing it even further and eliminate spikes. I cannot tell you how to know the exact value. You will have to experiment as it will be different for every IC I guess. Hell you can almost get a NON Ripple with enough capacitance.

But going LOW on capacitance isn't the solution.

EDIT: of course the layout of any switching regulator is playing a crucial role too
Think about it...

I don't see how a cap between Vout and FB would change the startup spike....


I DO see how a cap between FB and ground would help though. As the IC turns on, the cap on the feedback pin would prevent the voltage from rising too fast, and also help prevent any oscillations too. That's one of the things I had to do with my driver; I added a 0.1 uF cap from FB to ground. Before the driver was unstable, would oscillate and would always output a very high current regardless of the set resistor. With the cap, it's stable as can be.

I don't know man,

As far as I know a cap between FB and ground will result in oscillations on startup....

Well the way I see it too big of a cap on fb/gnd would keep it from rising very fast on startup. That would make the IC raise the current higher than it should be to try to raise the FB voltage. This equals startup spike and probably some oscillations too. What I don't know is how much would be too much and how much is needed to keep it stable. I'm going to try 0.1uf on the benboost along with 22uf input and output caps.