Voltage Controlled Voltage Amplifier/ Mr. Al

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
simf14
Posts: 24
Joined: Sun Jul 29, 2007 2:00 pm
Contact:

Post by simf14 » Mon Aug 20, 2007 10:41 pm

Wow what a great community there is on this board-everyone truly is interested in a lowly beginner and so helpful. This is a little embarrassing but I noted that with a moderate airflow over the heatsinks they now stay below 95 degree F on the voltage controlled voltage amplifier (12V, -12V supply, 5V control at 20mA in; 8V, 1.5 amps out.) Does this make sense for this project and the heat dissipation involved? I know Mr. Al might read through this thread in the am. I can't help but think I have some facts wrong with the solenoids. If there was some "short" in the solenoids how could I check for that and could it cause the 338's to heat up more than our theoretical calculations? These are big solenoids, bigger than of a roll of silver dollars. (max current 3.5A, power consumption 18W, coil resistance 1.8 ohms, 7.5Mhz coil inductance.) :cool: Also just throwing this out but I am using a switching regulated power supply, the switching should have no effect correct (?)

simf14
Posts: 24
Joined: Sun Jul 29, 2007 2:00 pm
Contact:

Post by simf14 » Mon Aug 20, 2007 10:44 pm

Yeah Chris talking op temps of 95F to 140F. (Nice pooch.)

Man I hope Mr. Al doesn't kill me. I have been assuming you guys have been talking F instead of C the whole time. Looking at Mr. Al's theoretical calculations now everything is actually running cooler than expected!!!! (emoticon for extreme embarrassment here) In my defense they confused meters and feet on some space shot right? Again thank you everyone, I just couldn't image it could run so "hot" to a "newbie."

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Post by Chris Smith » Mon Aug 20, 2007 10:55 pm

Your perfectly fine in those F ranges.

You say relays, but as stated earlier by some one else, most aren’t that sensitive.

Actual voltages can vary quite a bit and most have a min and max rating, and because of the swing in voltage and the hi resistance values, they usually can swing any where from 50% less, to in some cases 100% more. [not a guarantee there, but higher in most cases]

Heat from the unit can also tell if its too high, also if it over reacts way to fast, perhaps 100% over is too high.

Id check the R values on yours and see. If you have a sheet on it also see.

Unless they are precise, most can handle a over voltage.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Post by MrAl » Tue Aug 21, 2007 1:25 am

Hi again,


simf:
Keep in mind that when you test this power supply you cant just
switch it on and see if it gets hot. You have to set the input
voltage such that you get 6 volts output and you have the max
current of 3 amps flowing (or 3.5 amps if that's the true max).
Doing this is necessary because the max temperature rise is
what you are interested in measuring, and you can't get that
without getting the LM338 to have to handle the most power
it will ever see. For 12v input the max power point occurs
at 6 volts output (and solenoid drawing 3 amps). For 10v
input the max power point occurs at 5 volts output. Note the
max power point occurs when the output is at one-half of the
input voltage supply.
Since you also need to make sure the output is putting out the
full 3 amps, you might have to connect a resistor or two to the
output in order to make the output truely put out 3 amps if the
solenoid doesnt happen to be drawing that much current at that
point. In other words, to test this properly you need to set
the output voltage to 1/2 of the input supply (6v with 12v in)
and set the output current to 3 amps (or 3.5 amps if that's max).

After setting this test up properly like that, you can then
measure the case temperature. The best way to measure the
case temperature is to drill a small hole in the heatsink
just under the center of the LM case (or in the center of
where the chip is located within the case) and insert the
thermal probe so that it is touching the underside of the
case. I realize you may not want to do this however, so
lets say you get a reasonably accurate reading anyway.
You can add 10 degrees C to your reading to approximately
compensate, then add another 18 degrees to compensate for
the case temperature rise over the junction temperature.
Since the max junction temperature is 125 deg C and
you should stay 15 deg C under that spec and you have to
subtract 18 deg C and another 10 deg C, that leaves us
at 92 deg C case temperature max at the max power point.

Taking a look at the heatsink in that drawing, first it doesnt
seem like it is bolted down properly. I'd check that first.
The screws have to be tightened fairly well, and thermal
paste is a must but i guess you are using that already.
Also, the style and size seem to suggest it is only good up
to about 9 watts, meaning it could not handle the max power
of your circuit (3A times 6v, or 18 watts) because that is twice
the rating of the heatsink. Since forced air can increase
the cooling ability of a heatsink by over 2 times, using a fan
might get you there as then the heatsink might be able to
handle the full 18 watts.

Dont forget that the true test is when the LM device is handling
the max power that it will ever see, which is 18 watts. This means
6 volts output and 3 amps output (with 12v input). Reducing input
to 10v means only 15 watts max, but that sometimes works just enough.

Also, yes all these readings and calculations are done in degrees C
(not degrees F) because all the spec sheets use those units.
The conversion for deg F to deg C is:
degC=(degF-32)*5/9
and for C to F:
degF=degC*9/5+32

Since the max temp we calculated above was 92 deg C, this would
mean the max external case temperature you should measure is
198 degrees F.

Also note that if you expect a relatively large ambient temperature
rise in the field, you'll need to keep the test temperature lower
assuming you are testing at around 20 to 30 degrees C ambient.
For example, if you expect a 30 degree C rise in ambient in the field,
you'll need the test to confirm that the max case temperature is
about 152 degrees F.
LEDs vs Bulbs, LEDs are winning.

rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Post by rshayes » Tue Aug 21, 2007 4:05 am

One measure that might alleviate the thermal problem is to place a dropping resistor in series with the 12 volt supply to the circuit board.

The maximum output voltage and current is 6 volts and 3 amps. The regulator dropout voltage is 3 volts, so the minimum input voltage should be 9 volts. A 12 volt supply would allow as much as 3 volts drop across a series resistor. Since the maximum current is 3 amps, the resistor should be 1 ohm or less. With a 3 amp current, the resistor dissipation would be 9 watts and the LM338 dissipation would be reduced to 9 watts. This should halve the temperature rise on the LM338.

This will also lower the supply voltage for the op amp. The adjustment terminal of the LM338 is at least 1.2 volts below the output of the LM338. This will make it more than 4.2 volts below the op amp supply, so the op amp should work properly.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Post by MrAl » Tue Aug 21, 2007 4:57 am

rshayes wrote:One measure that might alleviate the thermal problem is to place a dropping resistor in series with the 12 volt supply to the circuit board.

The maximum output voltage and current is 6 volts and 3 amps. The regulator dropout voltage is 3 volts, so the minimum input voltage should be 9 volts. A 12 volt supply would allow as much as 3 volts drop across a series resistor. Since the maximum current is 3 amps, the resistor should be 1 ohm or less. With a 3 amp current, the resistor dissipation would be 9 watts and the LM338 dissipation would be reduced to 9 watts. This should halve the temperature rise on the LM338.

This will also lower the supply voltage for the op amp. The adjustment terminal of the LM338 is at least 1.2 volts below the output of the LM338. This will make it more than 4.2 volts below the op amp supply, so the op amp should work properly.
Hi rshayes,

I thought the dropout voltage of that regulator was 2 volts,
and the max output he wanted was 8 volts at 3 amps?
Ok, at 3 amps it's a little higher: 2.25 volts.
That's why earlier in this thread i said 10 volts would be the
min input, but yes 11 volts would be even better.
LEDs vs Bulbs, LEDs are winning.

rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Post by rshayes » Tue Aug 21, 2007 5:25 am

Hello Al,

8 volts out with a dropout voltage of 2.25 would require 10.25 volts in. With 3 amps output, the resistor would be .56 ohms. This would reduce the regulator dissipation by 1.68 watts. Not nearly as much, but sometimes every little bit helps.

Three terminal regulators are a little tricky. They have an internal thermal shutdown, but it might not be tightly controlled. I think it occurs when the junction temperature is about 150 C. Sometimes the junction to case thermal resistance is not very clearly stated either, making it difficult to predict the conditions where thermal shutdown will occur with a given heatsink.

simf14
Posts: 24
Joined: Sun Jul 29, 2007 2:00 pm
Contact:

Post by simf14 » Tue Aug 21, 2007 7:28 am

Could anyone suggest a very nice TO-3 heatsink that they approve of?

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Post by Chris Smith » Tue Aug 21, 2007 9:27 am

The ones I have used in the past look like a CPU heat sink, about two cubic inches and lots of fins. With a fan they easily cool down. Try All electronics, they usually have them.

Ill see if I can find a part number, also mouser and a few others have them but their mins are higher.

All Electronics is out.

A old small cpu cooler modified or perhaps Mouser?

A 233 mhz was small.

My PIC wont pop up on the specs, but the dimensions read about right at 45 mm or 2 inches cubed, To3 style, but at mouser they are expensive at almost 5 bucks each.

http://www.mouser.com/catalog/631/1702.pdf

Figure "E" is the biggest heat sink Im sure

I had some CPU to T03 styles, they really took the heat.

If you can machine some, you can find them at any surplus computer place. A lot better.

http://www.mouser.com/search/ProductDet ... -568303B00

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Post by MrAl » Tue Aug 21, 2007 12:17 pm

Hi again,


Chris, i could not find a heat sink on those pages that would
work for his application. Do you have a specific part number?

I found an Aavid heat sink, very large, model #6500B, but can not
seem to find it for sale anywhere.
I found a big one at Digikey, but they want 10 dollars for it.

Keep in mind we really need 2 deg C/W or better for this project,
maybe push it to 2.5 deg C/W.
LEDs vs Bulbs, LEDs are winning.

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Post by Robert Reed » Tue Aug 21, 2007 2:38 pm

Sim
Apparently your wattage requirements were quite a bit higher than was apparent from your original post that I saw. When selecting a heat sink , you really have to go thru the numbers to do it in proper engineering fashion. heres the chain of events to consider:

1) maximum wattage that will be dissipated across the regulators junction.
2) the thermal resistance of junction to case.
3) the thermal resistance of case to heatsink. The manufactures of thermal grease and mica insulators will give you specs on this and good combos will usually run about 0.3 to 0.6 degrees rise per watt.
4) now the last consideration is the highest ambient temperature in which it will be operated in and the neccessary heatsink to dissipate the heat transfer previously calculated in to that into that ambient .

This can be all woked out by using the manu's spec sheet. Once you have worked out the required dissapation of the heat sink, then you can select the heat sink, and of course you will want to leave a minimum of 10 to 15 degrees head room for these calculations. I always prefer 25 degrees, but many times that amount will become unreal due to the ungodly size of the heat sink required for that comfortable margin.
There are many heat sink manufacturers of heat sinks, but the first one that comes to mind is Wakefeild - Google that up and you can download their whole catolog with all the heatsink info you need.
You have been getting good info here in regards to regulator input voltage "headroom", however one thing I haven't seen mention of - and that is in regards to line voltage sags (maybe 5 or 10 %) and for that reason I always like to see about 3 volts of headroom at that point and that would be referenced not to average DC, but the lowest valley voltage of the ripple present.

User avatar
haklesup
Posts: 3048
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Post by haklesup » Tue Aug 21, 2007 2:39 pm

Looking at one in Mouser that resembles the one in your photo and its thermal resistance is in the order of 8 to 10 (depending on fin height).

If I turn the page a few to the Wakefield products I see Mouser#567-641-A. That's the 641 series heatsink and from the datasheet. I can see the thermal resistance is less than 1 C/W (with a fan) would be plenty for your app. Mouser cost is less than $8.

http://www.wakefield.com/pdf/Extruded_Heat_Sink.pdf
Most of the HS on this datasheet will work as well but the one noted is easy to obtain.

Make sure to get one predrilled for TO-3 or you will be drilling your own holes. Looks like your PCB will fit onto the heatsink rather than the other way around.

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Post by Chris Smith » Tue Aug 21, 2007 4:31 pm

The part I came up with was the ..........532568303B00 E style T-03

http://www.mouser.com/catalog/631/1702.pdf

My favorite was the CPU type which they didn’t have, but theirs was the thickest metal with the most amount of fins, and fans were suggested if needed.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Post by MrAl » Wed Aug 22, 2007 2:55 am

Hi again,


Guys, he already has a heat sink that would probably work ok
after including a fan, so he doesnt need to buy another heatsink
that will also require a fan.

The 641 heatsink would work ok without a fan, but the other small
ones wont work without a fan for his app.

Chris, did you check the thermal resistance for that heat sink?
It's pretty high.
LEDs vs Bulbs, LEDs are winning.

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Post by Chris Smith » Wed Aug 22, 2007 6:02 pm

They rate it at normal air, or forced air.

Those usually work better but as I said they don’t have the CPU types that I always used, tons of fins, thin aluminum, 2 inch cube, and with a fan, they ran real cold on hot items.

My basis here was visual, rather than the tried and true from the past.

They are out there but I don’t have all the resources I used to have.

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 35 guests