High current digital pot

newplumber

Senior Member
Hi
I was curious if there is a better way to make a high current digital potentiometer
I search the forums but didn't show any luck
my schematic is just a thought and I didn't add caps etc... just basic
what i am trying to achieve is make a harbor freight 120V dc mig welder with way more control on
amps so I can weld thin metal ...its funny the welder gives you 2 switches high/low ... but for the price you can/can't go wrong
of course i know you all have way better ways but I'm trying to keep it simple
one thing that helped me control current for the welder is just attach a few more extension cords in a circle but
with my crazy thought is to control it with better adjustments with a picaxe
your miscalculated friend
mark
FWIW ...harbor freight welder = 98 dollars
newplumbers wrongfully high current potentiometer = $98,000.00 :)
 

Attachments

hippy

Technical Support
Staff member
I am not an analogue expert but I think what you need is a power MOSFET.
 

techElder

Well-known member
Mark, it looks like you are attempting to replace a control that is built into the welder. Right?

So, I would like to see a schematic of (at least) that part of the welder. What kind of power is consumed in the control circuitry? You need to size your resistance for power as well as value.

Having said that, it also looks like you are just adding 2,625 ohm to (or replacing) a potentiometer on the welder. Then you want to create a sequence of switch closures (via relay) to reduce the value of that resistance to change the output of the welder. Is you new control going to be paralleled or series connected?

If you knew what the voltages are in that control circuit and how isolated it is from the mains voltage, then you could eliminate the additional electronics and just use manual switches. (Yeah, where's the fun in that? :D)

Or, you could just add a potentiometer to the panel and turn the knob with your PICAXE educated fingers. :D Just sayin'.

ADDED: http://forum.driveonwood.com/t/modifying-the-harbor-freight-120v-welder-if-you-already-own-one/64
 
Last edited:

techElder

Well-known member
After looking at what I think you have, it appears that the current control is done by switching between two different taps on the transformer primary.

This makes the request for a schematic of the welder much more important, because I have no way to know where you are planning to put your new "digital potentiometer."
 

newplumber

Senior Member
@ hippy ...I did some bread board tests with using my schematic and using IRFZ48N mosfets instead of mech relays trying to control led brightness by
running power thru my digital pot ...but using the mosfets wouldn't change the brightness (maybe I can draw my test of better view later)

@texas thanks for the "ADDED" and of course like always I don't explain important detail ...save it for you to guess :) but
for reals I thought the welder was switched to DC and now I know it is only AC .... back to the drawing board!
I have a few great wire feed welders that are DC with pot control for amps ...but before I throw this HF unit away
I am going to have some fun trying to make it worth while using so its not a dust collector.
I was planning on using my horrible digital pot in series ...with high amp rated resistors/relays connected from the ground clamp thru the digital pot then to the metal
of course I would make it so max amps would never exceed 20 and my welder would only be for welding thin metal like 16 gauge +
I mean what can go wrong?!! :)
 

hippy

Technical Support
Staff member
As I always admit; I am no expert with analogue electronics. But I do know power MOSFETs are used in high power audio amplifiers and so can be used to control high current in an analogue way rather than through using PWM or phase control. As to how and what circuit one would use; that is outside my skill set.

I am also not familiar with how welders work, voltages or currents they use. There may be the possibility of adding to the existing circuit or it may require a complete redesign of the innards.
 

newplumber

Senior Member
yes true hippy ...I will need to show alot more detail.
and for the digital pot idea even tho I probably wont use it in a welder
I am still going to see if it will work with mosfets or some small relays so
for this pot test I want to see if I can read ohms from 2550 to 10 in 10+/- ohm increments.


I think this welder project is balanced on dumpster/or lots of homework...
and I choose homework ...so when I open the welder box (UN plugged and caps drained)
I will let you know if its possible to picaxe it in simple ways
The first thing for sure is get rid of AC thru a full wave rec..for welding mild steel
 

techElder

Well-known member
I've worked on and built quite a few low voltage / high current units similar to what you are working on. 20 amps is not really trivial to control.

Still, the true schematic is necessary.
 

stan74

Senior Member
Does 110 V ac actually kill you. I think it's down to current. 240 in UK,usually kills you. Does ohms law have anything to do with this?
 

newplumber

Senior Member
LOL ya true Texas ...I seen many electricians touch live 120v ac with bare hands and tell me yes this outlet is hot/live...
but I like my fingers to stay pinkish white .

okay here is my schematic of a test and it didn't perform with results I was hoping for
of course what i gained from this test is to have mosfets moved from the drawer to the breadboard
and maybe back to the drawer ...I think I need to use some special relay or something to make this
hill billy digital pot work ...but its fun learning how not to make something :)
 

Attachments

techElder

Well-known member
Mark, for those mosfets to work, you have to be switching a resistor junction to a common line that is common to the mosfet SOURCE connection (ground!) So, your program would be to successively "short out" parts of the resistor string to adjust the total value. That may require you to change the value of the ladder of resistors.

Your choice of mosfet requires about 10 volts gate to source to get it to fully conduct. It probably won't be very active at less than 5 volts in any configuration.
 

newplumber

Senior Member
thanks texas I knew something was wrong but couldn't name the "something":) ... I am trying to find micro relays ...but all I get is solid state ones dc to dc...
which i don't think it would work for this application either...once I get relays for a test then I should be able to control
the digital pot with binary ...it will be interesting.
 

newplumber

Senior Member
Well texas its real secret ...because i don't even know what i am building :)
my idea was to make a digital pot using 74ch595 with micro relays and resistors
now since I found a small relay like this one..(or maybe something better)
https://www.alliedelec.com/m/d/1e9bfb051b841ae78908ad075b80bba3.pdf
az850-5
I am going to use two 74ch595 with resistors like 1,2,4,8,16,...
the whole project is learning ...my result at the end of the year should have a
digital pot from 0 ohms to 65,565 ohms .5W ..why? whats the point?
because I think it would be cool to build one from scratch
the resistors might be a pain to get but I will keep looking.
maybe its a horrible idea but I want to see if I can pull it off
 

hippy

Technical Support
Staff member
If we are moving away from the welder application and on to just a generic resistance box then I would say relays are probably a good option; just switch resistors in the chain in and out as required.

I would check on what the current ratings are for a 74HC595 but it should be easy enough to add ULN2803A Darlington drivers if required.

Getting resistors in non E-series values might prove difficult but you can probably daisy chain or parallel standard resistors to get what you need.

It would also be an opportunity to go further, perhaps have a rotary encoder and an LCD/OLED display to show what resistance has been selected.
 

hippy

Technical Support
Staff member
Using E12 series ...

1 = 1R
2 = 1R + 1R
4 = 12R | 12R | 12R
8 = 12R | ( 12R + 12R )
16 = 10R + ( 12R | 12R )
32 = 10R + 22R
64 = 27R + 27R + 10R
128 = 18R + ( 220R | 220R )
256 = 100R + 100R + 56R
512 = 12R + ( 1K | 1K )
1024 = 12R + 12R + 1K
2048 = 68R + 180R + 1K8
4096 = 3K9 + ( 220R | 1K8 )
8192 = 10R + ( 15K | 18K )
16384 = 15K + ( 1K5 | 18K )
32768 = 6K8 + ( 680K | 27K )
65536 = 68K | ( 820K + 1M )
 

AllyCat

Senior Member
Hi,

Note that that relay is rated to only 30 watts, 1 Amp switching and 50 milliohms "initial" contact resistance. So not too good for a Welder and 16 relays might give you almost one ohm before you start. :(

...should have a digital pot from 0 ohms to 65,565 ohms .5W .
Beware that a "variable resistor", a "potentiometer" and a "voltage divider", might do much the same thing but are fundamentally quite different. A potentiometer has three terminals and two resistive parts (from the wiper to each end), but of course you can use only one (part) if you wish. In some respects the variable resistor is actually the most difficult, for example, will you be able to find a set of resistors such that the 32.768k is one ohm more than the total of 1 + 2 + 4 + 8 + 16 + 32 + 64 + 128+ 256 + 512 + 1.024k + 2.048k + 4.096k + 8.192k + 16.384k ? (Hint: you won't). :)

An advantage of a "Pot" (Potentiometer = Potential Divider) is that if the resistance material (e.g. carbon or metal film) changes with temperature, then both parts change proportionally and any "divided" voltage won't change (by as much). However there is still the problem that "32.768 kohms" probably won't be 1 ohm more than "32.767 kohms".

But if you make a "Potential (i.e. voltage) Divider" using PWM with a low-pass filter, there is quite a good chance that "V * 32768 / 65536" will be V / 65536 larger than "V * 32767 / 65536".

Cheers, Alan.
 

newplumber

Senior Member
Hi sorry i was weekend vacation

@hippy thanks for the info and time saving table ...yes definitely on the LCD ..i bought a bunch of the cheap china ones...since the forums helped me get it to work.
("moving away from welder") yes I think maybe I will start a new thread for the welder if I still have a welder after i open the box and think i know what i am doing :)

@allycat... thanks Alan your kinda making my light at the end of the tunnel with a sound of a loud train...but thanks for explaining... I (simple Simon)
totally forgot about temperature change...but my heated garage stays somewhat stable in temp ...but again I in a little bit
see problems you mentioned ...I guess i was just thinking how easy it would be to PE code it...maybe I will start with just one
74ch595 and see if I can read ohms from 0-255 ...and see how close it can be in real world.
when I get my relays I will let you know in my uneducated guesses

@ the bear good thought and that is cool...i have to look into that thanks
 

AllyCat

Senior Member
Hi,

In the old days, the "E12" resistor series used to be called the "Preferred" resistance range, which is quite an appropriate name. Each value is about 20% larger than the previous, i.e. 10, 12, 15, 18, 22, 27, 33, 39, 47, 56, 68, 82, 100 etc. and can continue through many decades (i.e. up to 1 Mohm and above).

It dates from when resistors were made by mixing up a concoction with carbon powder, moulded, and baked in an oven. Then the value of each "resistor" was measured and put in a "bucket" appropriate to its value (give or take 10%). Now, with Carbon (or Metal) Film technology it's possible to manufacture whatever value is required to much higher accuracy, but the basic principle of "percentages" is still very relevant. Generally, you are likely to find that, for example, a 1.2 ohm resistor (which your switching box wouldn't have) to be more useful than 99 ohms (if it has 100 ohms anyway).

Now, there is the E24 series which "fills in the gaps", i.e. 11, 13, 16, 20, 24, 30, 36, 43, 51, 62, 75, 91 and even the E96 series of "1% tolerance" values. A "resistor box" which provides the Exx series is likely to be more useful than a linear range. Perhaps as many as E96 in the middle of the range (say 100 ohms - 100 kohms), but there are practical limitations with "accurate" values below 10 ohms or above 1 Mohms, so E12 or E6 values may be sufficient in those regions. Even the E3 series (10, 22, 47, 100) can be sufficient for capacitors and inductors, etc..

Of course you don't need to buy one resistor of every value, as hippy indicated earlier, it's possible to combine resistors in series or parallel (not quite as easy to calculate) for example 20k = 10k + 10k and 11k = 22k in parallel with 22k, etc. A good way to use the computational capabilities of a PICaxe. ;)

Cheers, Alan.
 

newplumber

Senior Member
Hi Alan another great piece of information ...okay so to get resistors 1ohm to 128 ohm for the use of using one 595 chip
I think I may have to parallel only a few to get within close range ...I never heard of E96 series ... just a thought since
it is "1%" tolerance so if I parallel say two 64ohm resistors to get 32 ohm would the tolerance drop to .5%? I have a lot of reading
to do still ...like example--> I build this best/worst digital pot box with a picaxe 20m2,74ch595,relays,resistors..etc..then when i add my LCD
16X2 I will still have to take forum night classes on how to read the ohms from the picaxe to display nicely on the screen.

Just a simple quick question to over/under educated professors here
if resistor ohms are changed by temperature ...then how does the computers on
planet mars rovers stay working in extreme weather changes? (you can say its over my head) :)
 

hippy

Technical Support
Staff member
it is "1%" tolerance so if I parallel say two 64ohm resistors to get 32 ohm would the tolerance drop to .5%?
If one has two perfect 100R resistors in parallel the resistance would be 50R. Using 100R +/-1% resistors could be ...

101R | 101R = 50.5
101R | 99R = 49.995, ~50.0
99R | 99R = 49.499, ~49.5

So that desired 50 can be anywhere between 49.5 and 50.5 which still works out at 50 +/-1%

if resistor ohms are changed by temperature ...then how does the computers on planet mars rovers stay working in extreme weather changes?
It is probably a factor of a number of things; designing so resistor changes don't have significant effect on operation, using resistors which are more immune to temperature change, keeping the internal temperature more constant.

A resistor NASA is using on a Mars Rover is likely a lot more expensive than what most people would be using.
 

newplumber

Senior Member
Thanks Hippy .. I think all my resistors are +/- 5% (gold) so i think I will order some 1% from digi-key which love the lighting shipping
and i am sure the designing (rover) is way more time/thought then anything else on earth. I think some day for kicks I am going to
apply at NASA and see what my chances are...maybe they don't know water runs down hill :)
 

hippy

Technical Support
Staff member
I am not sure if 1% are really much better than 5%, especially as resistor value increases. The top-end 32768R resistor could be out by 327R at 1% and 1638R at 5%.

If resistors were selected to be at or below their nominal values, and what values each had was stored, it would be possible to select the closest desired result. For example, wanting a 32768R might mean turning on that 32768R plus 220R plus 8R.

You won't have a full 65,536 selection of resistances, but you will possibly have more accuracy through the range.

Also how many 5% resistors would be outside what the 1% would deliver ? You might be better off with an accurate resistor marked 5% than an inaccurate resistor marked 1%.
 

AllyCat

Senior Member
Hi,

Yes, in general, combining resistors in series or parallel won't affect their overall tolerance very much.

Again, "in the old days", when resistors were manufactured, measured and then marked accordingly, those that were very close to a "preferred" value would be marked as close tolerance (perhaps 2% with a red band) and sold at a higher price. So if you bought an "ordinary" resistor (probably a 10%, silver band) there was a strong chance that it would NOT be the marked value but too high or too low by several percent.

But nowadays, when resistors can be manufactured to a specific value, there's a much better chance that you will get one close to the marked resistance. But still there must be some manufacturing tolerances (errors in measurement, etc.). If you take two similar resistors off the same "band", then they were probably manufactured by the same machine at the same time, so they will probably have similar errors. Therefore, you might be "unlucky" and get two resistors that are both near the edge of their tolerance band. But if you combine two resistors from different sources, there is a much lower risk that both will be "bad", so combining resistors may actually improve their overall tolerance.

If you needed, say, 11 kohms, there are several possibilities. Two 22k in parallel will give 11k but there's the risk they will have a similar tolerance offset, or 10k + 1K should give the same value, but the 10k will predominate. It will contribute most to the tolerance (so the 1k could be a cheapie) but also dissipate most of the power (heat), so combining resistors with similar values can be preferable. I can't think of any other combinations that would (theoretically) give exactly 11k, but several give just 100 ohms off, e.g. 2.7k + 8.2K = 10.9k, or 3.6k + 7.5k = 11.1k, etc., should such values be required.

Cheers, Alan.
 

hippy

Technical Support
Staff member
I suppose one question worth asking is what exactly you are trying to achieve.

As a 'for fun' or 'learning' project, a 'proof of concept' or 'novelty' it probably won't really matter what is used.

If you were going for something which could deliver a certified highly accurate binary selectable resistance you would probably be talking a very expensive construction with laser trimmed resistors and would also have to take into account relay contact resistances.
 

newplumber

Senior Member
Hi
My achievement is to have a LCD and show ohms starting at 0 ..(or with relay contacts close to 0) then run up to 255 just be reading ohms in
a resistor ladder... now after you pros gave me a lot of great information so with that, I plan on still building my relay potentiometer and see how close
at room temp I can have it count...I don't know much about resistors (except value and wattage) and its interesting to learn from yous. I started this project
to see if its possible to use in different applications...like example ...my welder but not possible for me ...or a older treadmill motor driver...which again seems
to much homework and plus there is way better ways for that in different forums on here...so overall I decided to just make a ohm reading display from my resistor ladder.
and of course its always for fun learning 30% of a ohm at a time. I don't own any 1% tolerance so it would be interesting for me to see if there is any if at all real change with 1% vrs 5% using resistors up to
128 ohm. And always thanks for the resistor info. simple question
If I have say 1/4 watt resistor at 100 ohm with a tolerance of 1% or even 5% will the resistor ohm value change using different wattage like
.125 watt = 99.5
.150 watt = 99.9 even if the ambient temperature doesn't change?

btw Alan I like your "in the old days" because for me it makes it clearer a lot of times how/why something is made better now days or current times

and hippy ..lol no I don't need no certified accuracy ... it way funner using dollar price items
 

premelec

Senior Member
Not sure what your question is... ideally a resistor has no change with temperature. In practice some have very small temperature coefficient of resistance and some pretty large... this latter type will then change quite a bit when dissipating more power and hence heating up. An interesting example of a large coefficient of resistance comes with the tungsten filament light bulb. About 8x change cold to bright temperature... ;-0 You can find practical coefficients for substances like coat hanger / baling wire [iron] copper wire [copper] on the Internet. Special alloys like constantan are close to zero change with temperature...
 
Top