Beginner Supply - DIY or Buy?

[clansd99]

Hi all,

I'm looking for my first linear bench supply and not sure what to buy/make. At the moment I am looking to make basic projects that build my understanding of transistors, op-amps, and feedback circuits. I imagine a 15V, 1A dual-channel supply would suffice but open to suggestions.

Being on a student budget, my limit is around $200. If a good supply can be found at that price point I'd prefer to buy one outright, but if not, my parts bin is well-stocked and this seems like a fun project. Unfortunately I haven't yet found a good guide on building your own supply that includes a safe approach to mains wiring. If anyone could point me to a few resources I would really appreciate it.

Thanks!
-Chris

P.S. the Siglent SPD3303X-E is very tempting but just so expensive!

[dobsonr741]

Power supplies are commodity. Would not build a single rail power supply. Dual/triple pushes the price up. Thus I have a Extech 382260, another nice option I considered was Global Specialties 1403. Again, there are so much to choose from, depending how much you willing to give in and buy cheap Chinese off of Amazon.

For op-amp dual rails experiments I use a TCA962 + 7812 + 7912 based breadboard supply PCB from the "Breadboard Friends" kit by Modularaddict. I do not consider this a DIY power supply. By the way, the modern analog rails are shrinking, down to +/- 2.5 to +/- 5V, so I consider +/- 15V an archaic convention.

[themadhippy]

could stick with the classics,lm317 and 337 and a handful of capacitors and diodes,if you make it variable then consider how your going to monitor the output voltage so you dont shove to much of it up yer project,tip -constantly having to check  it with your multimeter soon becomes a pita. A good compromise is  to make it switch selectable,a 2 pole 6 way switch gives you a range of useful  fixed voltages,you could  also make 1 position a varible so best of both worlds. 

[ledtester]

Consider getting a used "analog trainer" like the ones made by Elenco. For instance, the Elenco XK-525 has adjustable +/- supplies up to 15V , fixed +/- 12V and +5V supplies as well as a XR2206 function generator. The supplies are good for around 1A.

There is no current limiting, but I think you can get by without it for experimenting with op-amps and transistors.

Example ebay listing:

https://www.ebay.com/itm/175505151028

They are also easy to fix as you can get the schematics and they just use LM78xx and LM79xx series regulators.

[clansd99]

Not sure how to go about the mains wiring if I make my own LM317 circuit. Any links for that?
Dobsonr471, the supply you mentioned is a switchmode - how are the noise levels? I have a switching supply I got off Amazon but it has horrendous switching noise - 1.6V pk-pk at 9V output under light load!

The Elenco models seem like a good option although it can be hard to find (affordable) shipping to Canada.

[Martian Tech]

You should be able to find a good used Agilent E3620A (dual output, 0-25V, 0-1A) within your budget.

[pqass]

Not sure how to go about the mains wiring if I make my own LM317 circuit. Any links for that?

Easy dual bench PS: https://sound-au.com/project44.htm
This one includes current limiting: https://sound-au.com/project223.htm
PS topics (including safety and DIY or buy question): https://sound-au.com/articles.htm#psud
Other PS projects: https://sound-au.com/p-cat.htm#psu

Just study the whole https://sound-au.com/ site.
It has many practical projects and well written articles explaining analog electronics.

[clansd99]

Not sure how to go about the mains wiring if I make my own LM317 circuit. Any links for that?

Easy dual bench PS: https://sound-au.com/project44.htm
This one includes current limiting: https://sound-au.com/project223.htm
PS topics (including safety and DIY or buy question): https://sound-au.com/articles.htm#psud
Other PS projects: https://sound-au.com/p-cat.htm#psu

Just study the whole https://sound-au.com/ site.
It has many practical projects and well written articles explaining analog electronics.

Thank you! This is exactly what I was looking for. Will read through this 

[clansd99]

Wondering if there might be a fault with my switching supply. Picked up this switching supply on Amazon recently, but it has 1.6V peak-to-peak ripple/noise at 0.5A! Surely this can't be normal? I made this thread in large part to try and find a supply which wouldn't screw with my tests, but perhaps a refund is in order from the seller?

https://www.amazon.ca/NICE-POWER-Variable-Adjustable-Regulated-Higher-Precision/dp/B08C7PQRPQ/ref=cm_cr_arp_d_product_top?ie=UTF8&th=1

[pqass]

Wondering if there might be a fault with my switching supply. Picked up this switching supply on Amazon recently, but it has 1.6V peak-to-peak ripple/noise at 0.5A! Surely this can't be normal? I made this thread in large part to try and find a supply which wouldn't screw with my tests, but perhaps a refund is in order from the seller?

https://www.amazon.ca/NICE-POWER-Variable-Adjustable-Regulated-Higher-Precision/dp/B08C7PQRPQ/ref=cm_cr_arp_d_product_top?ie=UTF8&th=1
(Attachment Link)

It shouldn't be that bad.    Are you probing correctly? 
'Scope probe ground clip on minus (black-ring) and the sharp end on plus (red-ring) terminals?  Ignore the earth (green-ring) terminal.
Do you have the X10 switch on the probe in the correct position?  (your scope says its 1X)

[clansd99]

Yes, probing correctly as far as I can tell.
Scope probe ground clip on the black ring and the sharp end on the red terminal. Earth terminal (green) ignored.
I've double checked the probe is on X1, and switched out to try different probes/channels just incase it's a problem with the probe/scope.

[EPAIII]

I have several DIY power supplies. They use either the 78xx or 79xx chips or the 723 which allows adjustable outputs. I have used them for decades now and they have been perfectly adequate for most of the work I have done. The one I have used the most is a four output that uses only one transformer. The two positive outputs use 723s and are adjustable so I can have from around 2 V to a bit over +15 V. The two negative outputs are -5/6 V and -12 V and use 7905 and 7912 chips respectively. The -5/6 V is switch selectable by adding a pair of resistors in the ground connection of the 7905. I did this for designs where I wanted a +/- 6 Volt supply. These ICs provide short circuit protection so accidents have not caused any damage over the years. The balanced outputs have been handy for op amp circuits that I have designed a lot of.

Building linear, regulated supplies is not difficult. I just followed the manufacturer's data sheet recommendations.

But, today you can purchase a lab style supply for a very reasonable price. It will have both Voltage and current metering and all the protections. I could not build a supply, even without the meters, for the price of an already built one. I do have a nice one that cost about $50 USD and will probably buy more as time goes by.

You mention a safe approach to mains wiring. Generally speaking that is not that difficult. My present approach is along these lines:

1. You can purchase power entry modules that will have a fuse holder built in. The wall cord will plug into them. This is a very safe way of getting the power into your chassis. They come in a variety of wiring styles: I usually get the ones with solder lugs.

https://www.mouser.com/c/connectors/power-connectors/ac-power-entry-modules/?fused=Yes&termination%20style=Solder%20Lug~~Solder%20Terminal&rp=connectors%2Fpower-connectors%2Fac-power-entry-modules%7C~Termination%20Style

2. The power cords are available with the mating connector already installed. They are commonly called D-pin power cords.

3. I use a completely enclosed, metal (aluminum) chassis/box to enclose the power supply. Aluminum boxes or chassis are commonly available and easy to work with (as opposed to steel). DO use a metal box because it forms the primary protection against accidental shocks. The mains wiring is entirely inside the box except for the power cord and if you use a pre-assembled cord it will meet all necessary safety standards.

4. Good wiring practice from the power entry module to the power switch and transformer should be followed. This is normally point to point soldered wires. Although not normally required, heat shrink tubing can be used to insulate the solder connections.

5. The safety ground is connected from the power entry module to the chassis with a short ground wire. I normally use a crimped ring style terminal to one of the mounting screws for the power entry module. A star style lock washer should be used between the screw/nut and the terminal ring. That provides the return path to trip the circuit breaker in the building's breaker panel in case of a short to the chassis/box. This is a very important step.

6. For most of the power supplies I have built a 1 Amp, fast blow fuse has been more than enough. A 12 V, 5 Amp supply would be rated at 60 Watts and at 115 V that would be only about 0.52 Amps so that 1 Amp fuse value is OK. The fuse is installed in the power entry module. However, if your supply is for more than 100 Watts of so, you should actually calculate the fuse size needed.

That is pretty much it. I like to mount the transformer near the power entry module. The power switch is usually at the front. One of the transformer leads can usually be soldered directly to the power entry module. The other one often must be extended to reach the power switch. Heat shrink tubing can be used over an inline splice. I like to use a double layer of the heat shrink tubing (two sizes, one over the other).

[pqass]

You should be using X10 (switch on probe and likely need to tell the scope it's on X10 too) position most of the time.  Make sure you've increased the current limit on the PS so it doesn't go into current-limiting mode and put a small load on.  Increasing the load should increase the ripple.

Try to measure with a multimeter on lowest ACV setting.  It will return ACrms (0.56Vrms = 1.6Vp-p  see: https://www.allaboutcircuits.com/tools/rms-voltage-calculator/)
This should confirm the 'scope reading or not.

For comparison, I have a server-class 600W switching supply here where I added a 4A load on the 12V rail and get about ~40mVrms ripple (as reported by my bench meter).

[jasonRF]

Not sure how to go about the mains wiring if I make my own LM317 circuit. Any links for that?

Easy dual bench PS: https://sound-au.com/project44.htm
This one includes current limiting: https://sound-au.com/project223.htm
PS topics (including safety and DIY or buy question): https://sound-au.com/articles.htm#psud
Other PS projects: https://sound-au.com/p-cat.htm#psu

Just study the whole https://sound-au.com/ site.
It has many practical projects and well written articles explaining analog electronics.

Thank you! This is exactly what I was looking for. Will read through this 

I agree with pquass that the ESP site is great.  Awhile ago I was planning on building a basic, dual power supply based on the LM317/LM337, but ended up purchasing one already built off of ebay that is basically just a LM317/LM337/7805 triple supply right out of the datasheets.  It was $45 delivered so was cheaper than the parts (xformer+enclosure+heat-sink+pots+knobs+mains plug + switches+fuse holder+....), plus it looks okay.  It is similar to
https://www.ebay.com/itm/144871988063
and for the basic stuff you are interested in these simple supplies work fine, and if they are designed and built okay they will be reasonably quiet.

Of course for a little more money you might get a more capable supply on ebay with adjustable current limiting, constant current source output option, nice meters, etc. 

jason

[clansd99]

Starting to think there's an issue with my test setup.
Multimeter on mV AC range is showing 0.00mV across the terminals of the PSU. Scope reading of the peak-to-peak voltage is inconsistent, I'm now getting around 0.8V. Maybe a grounding issue?

My setup:
-Scope plugged into power bar
-Probe plugged into channel 1 and set 10X mode with 10X set on channel 1 in the scope
-Probe alligator clip attached to PSU black terminal through a wire about 3 inches long
-Probe springy hook attached to PSU red terminal through a wire about 3 inches long
-PSU black terminal connected to resistors through a long wire before feeding back to red terminal

Maybe it's some sort of radio interference? Or maybe the PSU and scope should not be on the same power bar?

You should be using X10 (switch on probe and likely need to tell the scope it's on X10 too) position most of the time.  Make sure you've increased the current limit on the PS so it doesn't go into current-limiting mode and put a small load on.  Increasing the load should increase the ripple.

Try to measure with a multimeter on lowest ACV setting.  It will return ACrms (0.56Vrms = 1.6Vp-p  see: https://www.allaboutcircuits.com/tools/rms-voltage-calculator/)
This should confirm the 'scope reading or not.

[pqass]

Starting to think there's an issue with my test setup.
Multimeter on mV AC range is showing 0.00mV across the terminals of the PSU. Scope reading of the peak-to-peak voltage is inconsistent, I'm now getting around 0.8V. Maybe a grounding issue?

My setup:
-Scope plugged into power bar
-Probe plugged into channel 1 and set 10X mode with 10X set on channel 1 in the scope
-Probe alligator clip attached to PSU black terminal through a wire about 3 inches long
-Probe springy hook attached to PSU red terminal through a wire about 3 inches long
-PSU black terminal connected to resistors through a long wire before feeding back to red terminal

Maybe it's some sort of radio interference? Or maybe the PSU and scope should not be on the same power bar?

You should be using X10 (switch on probe and likely need to tell the scope it's on X10 too) position most of the time.  Make sure you've increased the current limit on the PS so it doesn't go into current-limiting mode and put a small load on.  Increasing the load should increase the ripple.

Try to measure with a multimeter on lowest ACV setting.  It will return ACrms (0.56Vrms = 1.6Vp-p  see: https://www.allaboutcircuits.com/tools/rms-voltage-calculator/)
This should confirm the 'scope reading or not.

Looks fine to me.  3" wires shouldn't be a problem. I just remembered Dave's video "EEVblog #594 - How To Measure Power Supply Ripple & Noise" - he uses a specific adapter to minimize the terminals-to-probe connection. See 15:00 mark.  It's way past my bedtime. Goodnight.
Make sure there is no wire/shorting bar between green and black PS terminals.


Oh, don't forget to turn on [20MHz] bandwidth limiting on the 'scope.  That's the proper way to measure ripple.  Looks like you're picking up high-freq. crap and you're zoomed in.

And I was wrong about using the X10 setting for measuring ripple [specifically]. The X1 setting lowers the 'scope bandwidth and loads down your circuit (https://www.quora.com/What-is-the-difference-between-a-1X-and-a-10X-oscilloscope-probe/answer/Lawrence-Stewart-1).
See below.

[rdl]

I think you should start by building a basic LM317 power supply. That's what I did when I first got interested in electronics. I just followed the schematic given a Forrest Mims book. You can use a wall wart or a laptop power brick if you don't want to mess with the mains side circuitry, though it's not that hard. I'd also keep an eye out for a CPS250 on ebay or wherever. I see them all the time for $100-150 USD. There's a BK Precision version also, but their colors are a bit weird. One of these will pretty much cover all a hobbyist's needs.

[Calvin]

Hi,

P.S. the Siglent SPD3303X-E is very tempting but just so expensive!

why does seemingly everyone just look at those programmable 3303X-versions, instead looking at the considerably cheaper 3303C?
Who as a beginner needs a programmable supply anyway?
I bought the 3303C because I wanted something new, something linear and simply because of its easier to read display and easy handling -though with an rotary encoder and buttons it´s almost on the edge to 
But there´s nothing to say against a used linear supply in good working condition.

regards
Calvin

[MikeK]

My answer is to DIY and buy.  You can get the popular Korad supply for much less than your $200 budget and use the rest to build your own.  You'll learn a lot by building your own and you will have a bought supply without waiting for your DIY supply to be finished.  Even most hobbyists have more than one supply.

[Vincent]

I think you should start by building a basic LM317 power supply. That's what I did when I first got interested in electronics. I just followed the schematic given a Forrest Mims book. You can use a wall wart or a laptop power brick if you don't want to mess with the mains side circuitry, though it's not that hard. I'd also keep an eye out for a CPS250 on ebay or wherever. I see them all the time for $100-150 USD. There's a BK Precision version also, but their colors are a bit weird. One of these will pretty much cover all a hobbyist's needs.

The BK Precision version (model 1651) is often listed for cheaper than the Tek CPS250, AND it goes higher in volts, AND the fixed 5V rail has double the current capability. So I'd skip the Tek CPS250 and look for the 1651 instead personally.

[clansd99]

In the end I decided to purchase a CPS250 on eBay for $150 USD. I think it will cover my needs fine for now, and if I need something more in the future I'll be better positioned to know what I need.
I also purchased the Siglent SDG2042X signal generator with an education discount as I'm having a hard time working around the limitations of my XR2206 kit. Very fortunate to have the money to afford one after a generous christmas gift, hopefully it pays off in educational value!

[bidrohini]

I think buying is a better option. DIY can be considered later. It will be also possible to open a ready-made power supply and follow its PCB design.

[cellularmitosis]

could stick with the classics,lm317 and 337 and a handful of capacitors and diodes,if you make it variable then consider how your going to monitor the output voltage so you dont shove to much of it up yer project,tip -constantly having to check  it with your multimeter soon becomes a pita. A good compromise is  to make it switch selectable,a 2 pole 6 way switch gives you a range of useful  fixed voltages,you could  also make 1 position a varible so best of both worlds.

I took a similar approach with a simple two-lm317’s-in-series supply: variable voltage control, 6-way switch for the current limit.  https://github.com/pepaslabs/DualLM317BenchSupply

[cellularmitosis]

My answer is to DIY and buy.  You can get the popular Korad supply for much less than your $200 budget and use the rest to build your own.  You'll learn a lot by building your own and you will have a bought supply without waiting for your DIY supply to be finished.  Even most hobbyists have more than one supply.

OP I would also encourage this approach.

Because your current requirements are low, you could base the supply around a 40VA 24v transformer, which are commonly used in HVAC and thus are readily available for under $20 on eBay.

An excellent resource is HP’s app note 90: the DC power supply handbook.  https://www.changpuak.ch/electronics/PowerSupply/HP_AN90B.pdf

You’ll tend to find that with bench supplies, you typically need either low noise, or lots of current, but rarely both.  So you eventually want a switching supply (10 amp models are cheap), as well as a linear supply.  I would buy the switcher and DIY the linear.

Cheers!

[dobsonr741]

Dobsonr471, the supply you mentioned is a switchmode - how are the noise levels? I have a switching supply I got off Amazon but it has horrendous switching noise - 1.6V pk-pk at 9V output under light load!

The switch mode usually not directly powering the bench circuits I build. I always use a local LDO where I need exact voltage and low ripple, like for audio. Or a local buck converter for the digital supply. The high freq impedance of the hookup wires to the bench supply are not acceptable in many cases anyway, if high bandwidth or precision analog or high frequency digital is used. 1MHz is already high.

The switch mode supply gives me the galvanic isolation and can hone in on the initial voltage I need for the LDO(s), and can set the bring-up current limit for the first turn-on. Before I bought the Extech supply I used switching power bricks, 5V or 12V, but they were limited in power and/or voltage. The Extech can go up to 39V or 5A. The switching supply can give the occasional high current without high dissipation I would need to deal with when using a mains transformer + series regulator.

[Doctorandus_P]

For a beginner on a tight budget DIY or a kit is a good start for a power supply.
The 0-30V 0.002 to 2A power supply often sold with the name "Hiland" is a nice start. It is by no means a perfect power supply, biggest issue is that the opamps are over volted, the original design ( https://americanradiohistory.com/Archive-Practical-Electronics/70s/Practical-Electronics-1978-10.pdf) uses the LM741 which can go upto 44V.

As I said, it's not perfect, but the EUR 8 price point for this kit gives it an excellent performance to price ratio. You do need to add a transformer, meters and a chunk of aluminimum ( depending on what kit you buy).

Power supplies are just like multimeters. You need several of them, and you can start with a simple one, buy a better one when you need it, and the simple one will still be useful. Heck, for those EUR 8 I say buy two or three, and experiment with it. Learn how it works, make modifications, swap the opamps for others, try to improve it. Such a kit is an excellent opportunity to learn about opamps, transistors, stability of control loops and lots of related topics.

You also do not need to go to 30V 2A. I would rather have two power supplies that go upto 20V 500mA, than a single one that goes to 30V 3A. 30V, 3A has been the "defacto standard" for lab power supplies for 30+ years, but electronics has changed a lot and is far less power hungry as it used to be.

You can also get some other power supplies, depending on your need. The linear Hiland is relatively noise free for analog experiments, for motor control and other high power stuff, a big SMPS is better suited. A Step-up to 120Vdc (Common on Ali) is nice if you want to examine what diodes and transistors do when stressed above their normal ratings. (Overvolting BJT's or MOSfets often puts them in "avalanche" mode, where they conduct current without getting destroyed as long as the current is limited. Some are even rated for their avalanche energy).

Then save up your money to buy a decent oscilloscope. I would not skimp on an oscilloscope. The cheap scopes are toys or have severe limitations (that you may not understand, and therefore hard to recognize). I recommend to buy either Rigol or Siglent and just ignore the rest.

[wizard69]

If you are getting into electronics I really believe that simple power supplies are valuable learning projects.   So Yes you should DIY a power supply or two, probably a fixed voltage supply using common 3 terminal regulators.    That doesn't mean you shouldn't buy a quality variable voltage supply.   However hold off until the budget can cover what you finally set your sites on.   I wouldn't rush to put good money into a supply until  you know what or where your interests will go.

What do I mean here - different interests may require a different supply, for example if you are doing audio work a higher voltage capable supply may be needed, tube circuits are generally another supply altogether.   If you are into robotics you may need a different supply yet again.

As for a DIY supply or supplies, the real value here is to learn how supplies work, learn building techniques that use more forgiving parts and develop some analog debug skills.   This is important as power supplies have high failure rates in the real world, so understanding them is critical, even if your interests are all digital.   This is why I suggest that beginners seriously consider DIY power supplies as some of their first projects related to electronics.   You will need one almost right away and one is never enough.   Back in my early days you built a modest 5VDC supply with a 3 terminal regulator almost by default.