MAX1044 Failure?

[Strategy]

I'm powering a little synth circuit (Anesthetic Audio's clone of the TR-808 kick drum) using a GGG charge pump PCB fitted with a MAX1044 and a few other parts. Has worked like a charm until one day kaput for reasons I can't identify. Tested it out and I get a nice ~+9V but where I should be getting -9V, I'm getting a +3.56 volts. Same reading on the charge pump pcb and confirmed on the 808 pcb as well.

Those 1044 chips are not cheap, is this kind of failure commonplace and is there anything I should consider to prevent? causes?

The circuit does not draw a ton of juice but I don't know specs...

Strategy

[armdnrdy]

Those 1044 chips are not cheap, is this kind of failure commonplace
and is there anything I should consider to prevent?
causes?

The reply to the first question: yes
The second question: Use a LT1054CN instead. Rated for16 volts max. Higher current output as well. I've never heard of one failing.
The third question: The MAX1044 has a very low voltage rating -10 volts. If there is a swing in voltage or a spike..it can fail.

Snarling Dogs used the Max1044 in the Mold Spore wha pedal with less than stellar results......they kept failing.

[ggedamed]

I have found that MAX1044 is very sensitive to variations of the input voltage in the voltage inverter circuit. When I used it in voltage doubler circuits it withstood 18V with no problem. Don't ask me why, it's just the way it is with my samples - in inverter circuits they just die as soon as the input voltage goes over 10V.

[italianguy63]

This may be a conspiracy theory, but, I have seen it implied that using the MAX1044 boosted ( connect pin 1 and 8 ) to raise the frequency to avoid audible whine in the circuit may help lead to their demise.  Has anyone else seen this?  Another alternative is TC962, Mouser part# 579-TC962EPA.  MC

[kaycee]

I stopped using the MAX a while back because i killed a couple and for the number of reports of them dying. The LT is more expensive for me, with two of them I've connected up wrong and got a mini hot plate as a result.  Yet, Letting them cool and connecting up right they've worked without a hitch for months, tough little bugger that chip.

[Strategy]

Thanks for all the feedback. I'm assuming the LT chip is not a dead ringer for the MAX1044 relative to just inserting it as an alternative in the GGG circuit?

I'm looking at the Madbean Road Rage and it lists the LT and several other chips as alternatives but doesn't (in my quick read) seem to indicate that any reconfiguration is needed in the circuit to accommodate the various chips (?)

http://www.madbeanpedals.com/projects/RoadRage/docs/RoadRage.pdf

Strategy

[italianguy63]

I'm not personally familiar with the LT chip, but I will look at it now as another option to the MAX.  But, in my understanding, the charge pump chips seem to work alike, and the main important factor is balancing the pump cap with the output cap (which the charge pump circuits already do).  So, I think the simple answer is-- I think the LT is a plug-and-play replacement.

MC

[armdnrdy]

So, I think the simple answer is-- I think the LT is a plug-and-play replacement.

Some things in life are not that simple. 

Best to read both data sheets.

To increase the 1054's frequency, you connect a 5-20pf cap across pins 2 and 7.

To increase the 1044's frequency, you connect the "boost" pin to +V.

Depending on your board layout...not exactly plug and play.

Another thing of note:
The manufacturers of charge pumps usually show small capacitors (10µf) in their example circuits.
The reasoning is a small cap looks better when board space comes to mind.
These are filter caps that can be as large as space allows. Larger cap=smoother ripple.

[italianguy63]

Strategy-- do you have the original charge pump schematic for review?

[armdnrdy]

I'm powering a little synth circuit (Anesthetic Audio's clone of the TR-808 kick drum) using a GGG charge pump PCB fitted with a MAX1044 and a few other parts.

I believe that Strategy is referring to this one:
http://www.generalguitargadgets.com/projects/24-power-supplies/138-charge-pumps
The 9V bipolar supply

[italianguy63]

OK-- it is extremely stripped down.  Again, as I understand it. As C1 and C2 are the same size.. he should be good.  The TC962 I mentioned works the same way.  I have seen extra filtering around the circuit in other applications (filter caps and chokes), but the basic circuit is the same.  The loop of connecting pin 1 to pin 8 turns up the frequency.  This is supposed to reduce auditory "noise" by raising the frequency.  But, as I questioned above I don't know if turning the boost on is necessarily a good thing (I just don't know enough about it).  I had also heard grounding pin 7 in some applications gets rid of noise.  I think he could try substituting the other chip #'s offered here and he should be OK.  But, I am NOT a EE.  MC

[italianguy63]

To add-- C1 and C2 are the "buckets" that pass the power through; I have almost always seen them at 10uf.  I have seen other electrolytic (filter) caps on the input voltage and the exiting voltage.  They also usually match in size -- 10uf or 47uf.  The circuit I am using I put a 2nd filter cap of 100uf on the exit isolated with a small resistor or inductor.  But, I have not fully tested it yet.  So, you can take it with a grain for now.

[italianguy63]

Sorry-- one more thought.  The sundrie parts on the front end are there to isolate any noise coming in from the ground plane of the input jack in the case of any voltage differences.  (Opps... just read that in the description-- it explains it).

[armdnrdy]

To add-- C1 and C2 are the "buckets" that pass the power through; I have almost always seen them at 10uf.  I have seen other electrolytic (filter) caps on the input voltage and the exiting voltage.  They also usually match in size -- 10uf or 47uf.  The circuit I am using I put a 2nd filter cap of 100uf on the exit isolated with a small resistor or inductor.  But, I have not fully tested it yet.  So, you can take it with a grain for now.

I have built and tested many charge pumps for scaled down bipolar supply. I always use larger caps than depicted in the data sheets.
Linear Technologies LT1054 data sheet has more application info: http://www.jameco.com/Jameco/Products/ProdDS/1728360.pdf


Charge pumps are designed to run at a frequency where voltage losses are at a minimum.
In many circuits that use charge pumps this frequency is not an issue and can be kept at the "default" setting.
This becomes an issue when used in an audio application. The fixed frequency of a most charge pumps fall right in the range of human hearing. This charging and discharging of the capacitor is relayed into the signal chain as a high pitched whine.
This is where it is necessary to increase the frequency.

Lower performance is the lesser of two evils. 

While you can't attain optimum performance with the frequency increased, in most cases you alleviate the whine.

This is one reason that the MAX1044 isn't the best charge pump for the voltages we work with. Even at 9 volts we are pushing it close to it's limits!

I like to have a bit more breathing room before part failure. 

[italianguy63]

Agreed! (That it is better not to design things on the edge of the component's limits.  Again, my original query if anyone has experience with the "boost" option causing the MAX chips to die.)

[ggedamed]

Actually, that's no reason to not use parts values from the datasheet. The manufacturer made way more tests than any of us here before choosing the values. Besides, did you have any problem with capacitors having datasheet values? I didn't.

[armdnrdy]

Actually, that's no reason to not use parts values from the datasheet. The manufacturer made way more tests than any of us here before choosing the values. Besides, did you have any problem with capacitors having datasheet values? I didn't.

When creating a bipolar supply with a charge pump, the capacitors connected to the charge pump input and output are storage/filter caps.
When you connect your "new" bipolar supply to a circuit, these same caps are filter caps.

It is common knowledge that the larger the filter cap, the smoother the ripple. I have read in charge pump data sheets "larger capacitors might be needed."


Take a read through this:
http://www.maximintegrated.com/app-notes/index.mvp/id/2027

[italianguy63]

OK I did some work looking at the data sheets and frequency these chips work at.  They vary, and the operation of pin 1 varies too; in some chips it is a frequency "booster" (MAX1044 & TC962).  In others it is a frequency regulator (LT1054 & ICL7660).  

Hopefully, I extracted the data right, but I found these frequency ranges:

LT1054, 25Khz typical.
ICL 7660, 10Khz.
MAX1044 (unboosted), 5Khz.
MAX1044 (boosted), 30Khz.
TC962 (unboosted), 12Khz.
TC962 (boosted), 24Khz.

Our ears operate up to 20Khz.  So frequencies above that are desireable for this application.

An Interesting point to note here is the MAX boost is times 6, and the TC962 boost is only times 2.

In my estimation -- ALL of these are direct replacements for each other PROVIDED BOOST IS NOT USED.  Meaning pin 1 is NOT connected to pin 8 in the circuit (in this application).

TC962 is a replacement for MAX1044 even if boost is employed.  Maybe a good idea considering the reported failures using the MAX chip?

The LT1054 looks to be the easiest "no brainer."  Just be sure to not hook up pin 1.  If your circuit has it, Just use a SIP socket and cut off leg 1.

Somebody feel free to check my work.  I don't want to lead anyone astray.  MC

[ggedamed]

Actually, that's no reason to not use parts values from the datasheet. The manufacturer made way more tests than any of us here before choosing the values. Besides, did you have any problem with capacitors having datasheet values? I didn't.

When creating a bipolar supply with a charge pump, the capacitors connected to the charge pump input and output are storage/filter caps.
When you connect your "new" bipolar supply to a circuit, these same caps are filter caps.

It is common knowledge that the larger the filter cap, the smoother the ripple. I have read in charge pump data sheets "larger capacitors might be needed."


Take a read through this:
http://www.maximintegrated.com/app-notes/index.mvp/id/2027

A solution should solve a problem. Maybe you do have a ripple problem with the datasheet circuit? In my circuits - and I seem to prefer excessive gain applications - there was never a discernible problem with the charge pump ripple.
Then, how much capacitance should be used? 100u? 1000u? 10000u? The ripple is even lower if one adds an inductance to the circuit. Where should one stop? Trying to improve something without clear goals it's not a sane/cost-effective aproach. It's this approach that leads audiophyles to mass insane amount of capacitance in their circuits.

As they say, "better is the enemy of good".

[italianguy63]

On this discussion... my OPINION would be.

Use small transfer caps (10uf) as there has to be some limit as to how much power can be transferred depending on frequency.  Then, if you are concerned with ripple/noise, ADD addtional filter caps and circuitry as needed.  That's what I am doing.

My .02

MC