ZTEX USB-FPGA Modules 1.15x and 1.15y: 215 and 860 MH/s FPGA Boards

[DeathAndTaxes]

Could it be a polarity issue?

Center pos vs center neg?

[1714231047]

1714231047

[1714231047]

1714231047

[1714231047]

1714231047

[ztex]

The "smoke" issue is a plain lie from manufacturer, I NEVER wrote that email.

I can publish the header if you can't remember in this mail.

And what is about the image. If a part is blown up in this way it prduces smoke.

[gusti]

Could it be a polarity issue?

Center pos vs center neg?

No it wasn´t, checked that on the PS too.

[ztex]

If I was ztex, I might consider giving you a 50% rebate on the first board (and explain the PSU requirements better if needed), but certainly no more than that. Id actually find that

The (internal) policy for accidentally destroyed devices is a discount of 20% to 30% for the replacement order because I do not want to earn to much from this kind of pit.

It is always possible that a device is destroyed by not being careful (usually shortcuts, ESD) or by not reading the specification. This also happened to me. But it happened in the same way to two boards.

[gusti]

If I was ztex, I might consider giving you a 50% rebate on the first board (and explain the PSU requirements better if needed), but certainly no more than that. Id actually find that

The (internal) policy for accidentally destroyed devices is a discount of 20% to 30% for the replacement order because I do not want to earn to much from this kind of pit.

It is always possible that a device is destroyed by not being careful (usually shortcuts, ESD) or by not reading the specification. This also happened to me. But it happened in the same way to two boards.

I think you need to design a better voltage regulating circuit, with some kind of protection.
Or make it clear not to connect consumer grade PS, only PC PSU with a better regulation.

[ztex]

You also reported that the "working" power supply was pushing 13.5V.

AFAIR 11.5 V under load and 13.5 V in idle. It seems to be an unregulated supply without bulk capacitors.  (With bulk capacitors the idle voltage would be higher.)

The voltages he measured are mean voltages. The peak voltages are drastically higher.

The one thing I would advise ztek to do is offer an optional power supply for sale.  If you already do then that makes gusti claim even weaker but if you don't you might want to offer one.  Users who may be concerned about their ability to verify a power supply before using it could just buy one from out.  Those of us who intend to power 20+ from a ATX PSU wouldn't need to.

Any power supply with a peak voltage less than 18V (recommended: 16V) and a minimum voltage of 4.5V can be used.

All stabilized supplies are save if their nominal voltage is within this limits. Switching supplies are always regulated.

There is a problem problem with selling power supplies:  different countries have different plugs and different voltages. Expensive international kits would have to be used and due to the size shipping gets even more expansive. Furthermore, most of these 1.15x FPGA Boards are used in clusters with a special power supply, see initial post of this thread.

[DeathAndTaxes]

There is a problem with selling power supplies:  different countries have different plugs and different voltages. Expensive international kits would have to be used and due to the size shipping gets even more expansive.

Yeah me being amero-centric again I forgot how many different voltage/plug/cycle combinations there are.

Furthermore, most of these 1.15x FPGA Boards are used in clusters with a special power supply, see initial post of this thread.

Agreed which is how I intend to use it.

[cicada]

There is a problem problem with selling power supplies:  different countries have different plugs and different voltages. Expensive international kits would have to be used and due to the size shipping gets even more expansive.

Those who are unsure about what PS to use, can't test voltages, etc etc, may still prefer the piece of mind buying an overpriced PS from you.

Don't bother selling 'kits', sell verified power supplies separately and allow the buyer to decide which type they need.  Even if you only stock power supplies for 60hz 120V US and 50hz 220V EURO you'll cover most of your bases.  Absolutely charge extra shipping, and absolutely mark up the price to cover your extra costs.

Out of 100 single-board orders you sell, maybe 10% of those buyers will know the difference between regulated and unregulated power supplies.  Cover your ass on the other 90% who will just try plugging in whatever fits in the socket.

[edit] While, yes, there are a large number of possible combinations if you wish to cover all countries, offering A, C, and G transformers will cover almost all of the major ones.

http://en.wikipedia.org/wiki/Mains_electricity_by_country

[fizzisist]

You also reported that the "working" power supply was pushing 13.5V.  That's 11% overvoltage.  Tell you what put 13.5V into an AMD video card and then when you see smoke try it on a second card.  Then try getting an RMA.  Be sure to tell them exactly what you did.

Most electronic warranties don't cover overvoltage.  

The one thing I would advise ztek to do is offer an optional power supply for sale.  If you already do then that makes gusti claim even weaker but if you don't you might want to offer one.  Users who may be concerned about their ability to verify a power supply before using it could just buy one from out.  Those of us who intend to power 20+ from a ATX PSU wouldn't need to.

Like I said in the other thread (now locked), that's the voltage when there is no load. I don't understand how voltage without current can do damage to the board. As soon as there is some current draw, it should drop to acceptable levels. If it had a huge output capacitor, then I could see it potentially doing some damage. But ztex, you said it probably didn't have one?

I'm only asking in the name of science here, not to slander ztex or his design or to take the side of the customer. I also want to make sure this type of thing doesn't happen to a customer using one of our boards. I've definitely lost some sleep worrying about something like this happening, and now that it's happened to someone else, I'm even more worried. If it turns out that using an unregulated "12 V" supply can really do this much damage, I don't want to ship a single board to a customer until I feel that our customers are sufficiently well informed not to do this.

Now, can someone tell me how one can tell that a supply is unregulated or not? Other than measuring it with a multimeter, of course. As far as I know, there is nothing on the label to indicate this.

[DeathAndTaxes]

There is a problem problem with selling power supplies:  different countries have different plugs and different voltages. Expensive international kits would have to be used and due to the size shipping gets even more expansive.

Those who are unsure about what PS to use, can't test voltages, etc etc, may still prefer the piece of mind buying an overpriced PS from you.

Don't bother selling 'kits', sell verified power supplies separately and allow the buyer to decide which type they need.  Even if you only stock power supplies for 60hz 120V US and 50hz 220V EURO you'll cover most of your bases.  Absolutely charge extra shipping, and absolutely mark up the price to cover your extra costs.

Yeah I agree selling kits would be worthless.  If power supplies are offered they should be separate and optional.

Another option would be to find a large retailer (amazon, digikey) that has a compatible PS. 
Ztek could test it an then provide a link to it from the product page.

[cicada]

Now, can someone tell me how one can tell that a supply is unregulated or not? Other than measuring it with a multimeter, of course. As far as I know, there is nothing on the label to indicate this.

There really isn't a reliable way to know outside of testing with a multimeter.   A well regulated 12V supply shouldn't spike above 13-14V unloaded, and most will read slightly below the rated voltage, somewhere around 10-11.5V.

Generally cheap wall-worts that come with other gadgets (switches, routers, etc etc) are unregulated, the regulation circuit is part of the device they're intended to power.  A bad unregulated 12V supply can often spike to 20V+ unloaded, and the same supply can dip well below 12V under moderate load.  This is often the bane of those beginning in microelectronics (amtel/arduinos/etc).

The only way to know for sure that the PSU the buyer is using is going to work and not fry the board is to supply it yourself.

[DeepBit]

Don't bother selling 'kits', sell verified power supplies separately and allow the buyer to decide which type they need.  Even if you only stock power supplies for 60hz 120V US and 50hz 220V EURO you'll cover most of your bases.  Absolutely charge extra shipping, and absolutely mark up the price to cover your extra costs.

While, yes, there are a large number of possible combinations if you wish to cover all countries, offering A, C, and G transformers will cover almost all of the major ones.

We aren't living in the past already :)
Most modern switching power supplies are designed for 100-240 range or even better (sometimes marked as "autovoltage"), this covers the entire world, only some cheap physical plug adapters may be needed.

Generally cheap wall-worts that come with other gadgets (switches, routers, etc etc) are unregulated, the regulation circuit is part of the device they're intended to power.

Well, there is was a regulator as the part of the device.

Like I said in the other thread (now locked), that's the voltage when there is no load. I don't understand how voltage without current can do damage to the board. As soon as there is some current draw, it should drop to acceptable levels. If it had a huge output capacitor, then I could see it potentially doing some damage. But ztex, you said it probably didn't have one?

I can't imagine how this damage could have happened if he used correct polarity and the board was good at arrival. I doubt that the PS was really that bad.
I can think of shorting something at the back of the board by stray screw laying on the table, but ztex says that this regulator is overcurrent-protected and that would be strange to fry both boards same way...

[DeathAndTaxes]

I can think of shorting something at the back of the board by stray screw laying on the table, but ztex says that this regulator is overcurrent-protected and that would be strange to fry both boards same way...

The weird thing is two board and destroyed the same way.

Maybe the fryer could post a picture of power supply label.  Reverse polarity was the first thing I thought of but the fryer claims it was proper polarity.

[ztex]

Like I said in the other thread (now locked), that's the voltage when there is no load. I don't understand how voltage without current can do damage to the board.

If an over voltage spike occurs, the switching regulators breaks through, i.e. you have the input voltage on the 1.2V and 3.3V rails. This immediately destroys most chips and capacitors.

As soon as there is some current draw, it should drop to acceptable levels. If it had a huge output capacitor, then I could see it potentially doing some damage. But ztex, you said it probably didn't have one?

A bulk capacitor is on the FPGA board. The job of input capacitors it to eliminate spikes (ceramic types) and to reduce slower peaks. Without that capacitors the board would be even more sensitive.

According to Gusti the damage occurred in idle mode.

Now, can someone tell me how one can tell that a supply is unregulated or not? Other than measuring it with a multimeter, of course. As far as I know, there is nothing on the label to indicate this.

This is usually stated on the label. If not, and it is a transformer based thing, you have to be careful. Switching regulators are always regulated.

[ztex]

Yeah I agree selling kits would be worthless.  If power supplies are offered they should be separate and optional.

Another option would be to find a large retailer (amazon, digikey) that has a compatible PS. 
Ztek could test it an then provide a link to it from the product page.

O.k. I will add power supplies for EU to the shop at begin of next year. There exist small variants which can shipped per airmail and which I can offer for about 10 EUR.

As I wrote, switching supplies are save because they are always regulated. I.e. in order to find one you just need to enter "switching power supply 12v" on ebay:
http://www.ebay.com/sch/i.html?_nkw=switching+power+supply+12v

They cost about $10. If I would offer them in my shop they would cost about $30 to $40 including additional shipping costs. (Because they either need to be imported (high shipping costs) or they I have to purchase expensive international variants)-

I will add the hint to purchase switching supplies from Ebay within the next days to http://www.ztex.de/usb-fpga-1/usb-fpga-1.15x.e.html#con5. The problem: almost no one reads it.

[sadpandatech]

I will add the hint to purchase switching supplies from Ebay within the next days to http://www.ztex.de/usb-fpga-1/usb-fpga-1.15x.e.html#con5. The problem: almost no one reads it.

  If nothing else, atleast some good came out of this experience so far.  Aye, people may not read it but it grants you the ability to point it out when they blow their junk up. ;p

[ztex]

I can think of shorting something at the back of the board by stray screw laying on the table, but ztex says that this regulator is overcurrent-protected and that would be strange to fry both boards same way...

Short circuits or other random errors like ESD or defect parts (with an unacceptable life time -- all boards are functionally tested) can be excluded since both board are damaged in the same way: both voltage regulators on both boards are destroyed. The damage on the 3.3V regulators is even visible.

[2112]

I'm only asking in the name of science here, not to slander ztex or his design or to take the side of the customer.

In the name of science I'm answering:

1) cheap multi-meters measure the average voltage
2) int sin(t),t=0..pi = 2
3) thus peak to average ratio for a full-wave rectifier is pi/2 ~= 1.570
4) 12V * pi/2 ~= 18.9V, 13.5V * pi/2 ~= 21.21V
5) unregulated but filtered power supply will charge its filtering capacitors to the peak voltage in (4)
6) a silicon device designed for 16V maximum input may release its magic smoke if supplied with the unfiltered rectified waveform with peaks as high as (4)
7) if the power supply was first plugged in to the mains side and then to the load side then the filtering capacitors will get partially discharged from the state (5) to their designed average
8) the safer way of powering up is first to plug in the load side of the supply to the device then plug in the supply to the mains outlet.

The above is what science says.

But the whole thread is an example why selling electronics to consumers demands sandbagging: both financial markups to cover losses caused by pissed of consumers and to safeguard devices from the accidental abuse by the uneducated consumers.

[gusti]

I can think of shorting something at the back of the board by stray screw laying on the table, but ztex says that this regulator is overcurrent-protected and that would be strange to fry both boards same way...

Short circuits or other random errors like ESD or defect parts (with an unacceptable life time -- all boards are functionally tested) can be excluded since both board are damaged in the same way: both voltage regulators on both boards are destroyed. The damage on the 3.3V regulators is even visible.

Sure, all is perfect designed and tested on your side, and the customer (me) is the stupid who cannot choose a proper power supply.
Customers, beware.

[ztex]

Sure, all is perfect designed and tested on your side, and the customer (me) is the stupid who cannot choose a proper power supply.

If you would have accepted the possibility that you choose an improper power supply you would have one running board and a 30% discount for a replacement order.