Bench Power Supply Best Practices

[cortes]

I'm just starting to build my electronic bench. For a power supply, I got a Techpower TP3005T, which is a single output. Then I went into a daze and when I came out of it, I had bought a Siglent oscilloscope and function generator. I was in full gear acquisition syndrome. I then set my sites on and Siglent SPD3303X. One of the features I like is the ability to turn off the power outputs. That got me thinking about best practices in working with a power supply.

With the TP3005T, I've been disconnecting the leads when making changes on my breadboard. If I turned off the power supply, I turn it back on and then connect the leads as I've read about power spikes on start up. The TP3005T has some protection for it, but I don't understand it fully. Were I to go with the SPD3303X or similar device, would turning off the power output for a given channel be sufficient to protect the circuit or should the leads always be disconnected? If the latter, I may just get another TP3005T given it's only $80US. I'm not at a point that I can use the more advanced features of the SPD3303X.

Here's an Amazon link for the TP3005T: https://www.amazon.com/Tekpower-TP3005T-Variable-Linear-Supply/dp/B00ZBCLJSY/ref=as_li_ss_tl?ie=UTF8&linkCode=sl1&tag=dcpsupplies-20&linkId=b61460804bc2873b0cd34aea912e1adf

[mawyatt]

We have 3 of the SDP3303X-E (all enhanced) and a GPP4323 and all have the ability to completely disengage the outputs, either all or individually. This is a baseline absolute must for lab use IMO. You need to be able to remove power from a circuit ASAP under distress conditions and this feature provides such. You will also appreciate the low noise levels on these linear supplies, they all are very low, and also like the stability of the set points and sensing.

Best,

[BravoV]

Don't know much on Techpower TP3005T, but think of this scenarios.

First ...

- You used the PSU, set it to 12 Volt, worked on it, then you disconnect the output wires to you breadboard, everything was fine n dandy.
- You turned off the PSU, then went to pee. 
- You returned back, while the PSU off, you hook up the output to your "other" breadboarded circuit that can handle "max volt" at 5 Volt.

Question ... will it zap your circuit with 12 Volt ? As the output of PSU, even at off position, its possible that its still has the residual 12 Volt at the output. 

Second ...

- Same as above ... after the pee step. 
- While the PSU is disconnected to load, you turned it on, set the voltage to "lower" the before, say at 5 Volt again, and then you confidently hook up to your 5 Volt circuit without measuring it 1st.

Question ... same as prev. one, if you did that in few seconds between 12 volt to 5 volt, will the residual 12 Volt still there, as I've witnessed a crappy bench PSU that took few seconds to settle to lower set voltage as it does not have the internal bleeding resistor at the output. 

Its much safer to learn and know your bench PSU well 1st, try measure and do some experiment before you toast your circuit. Most high end or well designed bench PSUs do not have this problem.

Yes, I'm a bench PSU aficionados, own > 5 adj. bench PSUs  , just look at my sig at below right corner of my post. 

[Shock]

Use your oscilloscope and measure it turning on during various scenarios loaded and unloaded using single shot or single trigger mode.

Daves probably got a video on it somewhere. The other one you will want to know is your power supplies noise and ripple.

[alligatorblues]

Choose your bench supply as you would the mother of your children! Do you want a cheap, unreliable woman whose full of unpleasant surprises that appear at the worst possible times? No, you want a quality mother for your children, who is reliable, respectable and predictable!

How would you feel if your teenage son found nude shots of his Mom online, but didn't realize who it was, and might be getting really turned on by her? That's how you'd feel if you walked up to your bench, smelled charred components, and realized you bench supply current-limiting failed! Think a bout that poor transformer!

I almost wanna cry! I did that once, but it was my fault! I was using a braid-clad temperature probe to monitor part of my setup. I had inadvertently turned the current-limit all the way up! The metal braid draped over 2 posts on the bus I was using, and shorted it. Dead short was going through the PSU transformer for 11 hours! But, it was a Lambda avionics DC supply, SO I just put a fan on it for a few hours, and it was OK.

*THAT'S* what you want for a bench supply! A professional item. Mine is 0-60V 0-15A, with programmable voltage cut-off, and totally bulletproof. I use it for my Peltier experiments, because it'll give me a lot of current! I got that for 105.00 on eBay. It weighs about 90lbs (41kg).

I was pretty sure it worked, because you'd have to do something so radical and evil to break it it  is almost impossible! I tested it just about up to its full 900W, and it was within < 1.0mV all the time! Another supply would have been toast in the above scenario!

I never purchase bench equipment new. I do with handheld DMMs, though! If you know what your doing, practically any instruments will do. I have a SR104 10k resistor from 1974!   

[bd139]

Know thy power supply well is the only rule. You seem to be aware of the limitations of yours which is a start.

A fine example is a lot of really high quality supplies have an electrolytic capacitor across the output after the output switch. You only find this out when you accidentally charge it up and then dump it across a diode which quickly becomes a deadode.

Also switching supplies generate a lot of noise. As long as you know this and what applications it can be used for it’s fine.

So understand what you have is important. If you’re going to buy something else I would consider important specifications:

1. Linear supply. Low noise.
2. Separate line and output switches.
3. Current limiting with fairly good resolution at the low end.
4. No horrible fans
5. Made of standard bits and has a service manual.
6. Optionally analogue controls. These allow you to smoothly sweep operating points rather than step through with an encoder. This is an underrated and very useful feature.

Possibly buy second hand. I personally like the TTi PL310 / PL320 / PL330 power supplies over all the really fancy ones. You can get a couple of those here for half the price of the Siglent crates.

[Kleinstein]

For most tests on the bench one does not need a lot of current. So a 2 A lab supply is usually large enough, though many start at some 3 A. Similar for the voltage, there is little need for more than 30 V and today even 20 V. Wort caue a failing (bad ouput transistors) linear supply may give out about 1.2 to 1.5 times the normal maximum voltage. So there is a safty plus if the supply is for 30 V or less.
For a cheap china model consider the rated current as a short time limit - they may overheat when when used at the full current and an unfavorable voltage (like a short) for a longer time. The full current at a high voltage may also lead to extra ripple.


Most lab supplies have a seprate enable function for the output and the idea behind this is to get a controlled turn on and avoid overshoot. Hewever this function is not allways implemented correctly. So it is a good idea to check the supply first (especially a cheap China model). If working OK the ouput enable function is normally enough during changes to the circuit. An for a good reason many supplies when turned on at the mains switch start with the output disabled. It depends on the circuit if it makes sense to additionally remove a cable - if you trust the supply and nothing super sensitive one can usually just use the enable.

[cortes]

Thanks for all the great information. I was originally looking at the classic choices of the Rigol 832 and the Siglent SPD3303x-e. I didn't care for the 3303 user interface. I like the one for the 832, but the unit's 41cm depth is an issue.

Then I made the mistake of looking at the Keysight EDU36311A. It's a lot more expensive, but when looking for the mother of your children you shouldn't skimp 😉 The depth is only 24cm and the user interface is a bit better than the 832 for my personal preference. It has a ripple and noise rating of <1mVrms/5mVpp for normal mode voltage. Then I looked at the 832 and it had a rating of <350μVrms/2mVpp for normal mode voltage. It seems like I would have to step up to the Keysight E36312A to get the same specs. Are the Keysight units really better? I would think so, but I'm not experienced enough to know how.

My use case is for the development of home automation devices using ESP32's and Arduino's.

[bd139]

The new keysight units should not be touched. Keysight will not support individual customers anymore. There are a few threads on the forum about it.

If you’re developing home automation stuff then anything will do. The keysight is overspending. Look at Korad units.

[cortes]

The new keysight units should not be touched. Keysight will not support individual customers anymore. There are a few threads on the forum about it.

If you’re developing home automation stuff then anything will do. The keysight is overspending. Look at Korad units.

That's good to know. I'll check it out.
Found a thread here: https://www.eevblog.com/forum/testgear/keysightcare-you-now-need-a-paid-subscription-to-ask-simple-questions/msg3729391/#msg3729391

[Wallace Gasiewicz]

Switching PS do make noise. They are designed so that noise does not affect the particular instrument they are in.
I did notice that you are working with low power devices but here are my comments:
I pick up PS at Hamfests second hand. Someone recently gave me am old small Lambda. Boy!, it is really nice!
For big power I use an old HP 6268 B, maybe 70 lbs.
I think that in a lot of instances a good multimeter check of the voltage of the PS is a good idea. Especially with the older PS with analog meters. This can save you a lot of trouble. If you use banana plugs for power, you can stack a multimeter lead on top of it for better voltage readings, when you think it is necessary.
A triple PS is very nice for most projects, again just be careful to measure the output volts vs the meter or digital display. If you are working with TTL voltages the 5 Volt separate output is very helpful. If you are just working with low power stuff, triple is probably all you need,
I sometimes use a big 100 ohm resistor (big, so I can hold it) to short the PS leads during set up. This way I can see if the ammeter is working and I can make sure that the voltage has been reset.
I almost always use Banana connectors for quick disconnect. Sometimes faster than getting to the switch.
You can get some banana connectors and some nice flexible black and red power wire and make your own jumpers with whatever alligator, hook or other connector you want.
If you are going to work on radios, like Ham radios,, a nice 14V PS is a must. A good brand name here is a good sign, Like Yaesu etc, but it is nice to have one with the meters in it.
The new switching PS for radios are quite nice but they do make some noise also, It is just that the particular noise they make does not usually affect these radios.  I prefer Linear PS.
You can recoup some test equip money by buying broken 14 V Linear PS for radios and fixing them, usually the problem is in the regulator board or regulating transistors and these can be easy to repair.

[cortes]

Thanks again for all the advice. I went with the Rigol 832

[wizard69]

I'm just starting to build my electronic bench. For a power supply, I got a Techpower TP3005T, which is a single output. Then I went into a daze and when I came out of it, I had bought a Siglent oscilloscope and function generator. I was in full gear acquisition syndrome. I then set my sites on and Siglent SPD3303X. One of the features I like is the ability to turn off the power outputs. That got me thinking about best practices in working with a power supply.

Your credit card must love you.   As for the TechPower see my comments about binding posts below.

As for best practices for power supplies I often take a different approach than many here, that probably due to years in automation and industrial electronics.   First let me point out that a variable voltage current limiting bench power supply can be very valuable on the bench.   In some cases that should be a supply with dual tracking outputs.   For some types of experimentation, design iteration testing and so forth they can be very useful. 

The problem is I see them less than ideal for a lot of what is done on a hobbyist bench these days.   I'd much rather have a fixed voltage supply that matches the required voltage requirements for the board being worked on.   In many cases this means 5VDC, 12 VDC or 24 VDC.   Such supplies can be had fairly cheap, new or used and just require the ability to build them into a suitable cabinet if they are not already user safe (wall wart or similar).   These can also be user made supplies.

So when you say best practices you really need to consider context.   If you are working with pre designed single board solutions, they will likely be designed to run on a specify voltage range and a fixed voltage supply is what you would want.   It is only when you start to work with your own designs, especially analog, that you really would benefit from a variable supply.

With the TP3005T, I've been disconnecting the leads when making changes on my breadboard. If I turned off the power supply, I turn it back on and then connect the leads as I've read about power spikes on start up. The TP3005T has some protection for it, but I don't understand it fully. Were I to go with the SPD3303X or similar device, would turning off the power output for a given channel be sufficient to protect the circuit or should the leads always be disconnected? If the latter, I may just get another TP3005T given it's only $80US. I'm not at a point that I can use the more advanced features of the SPD3303X.

This might start an argument but I believe more hardware has been trashed by people screwing up with variable power supply settings than any power up transients seen on a power supply output.   Beyond all of that it is a pretty poor board that can't accept transients to some extent on the power input.

As for advanced features on any instrument this is a double edge sword.   If you don't buy instruments with such features you never need to worry about using them.   One the other hand if the features are not there you can never learn to use them.   Obvious right?    The problem is many instruments are bought with the intent to learn to use the ""advanced"" features but the opportunity to apply them never arrives, so a lot of functionality ends up unused.   When it comes to power you cant get by with very crude power supplies perfectly fine, some of the first supplies I built where perfectly crude     These days I'd want better out of even a home built supply, simply having voltage and current displays are a huge benefit in my mind.   Those displays save trying to hook up a couple of meters and the clutter that creates.   However do most users leverage the USB or Ethernet connection - I kinda doubt it.

Here's an Amazon link for the TP3005T: https://www.amazon.com/Tekpower-TP3005T-Variable-Linear-Supply/dp/B00ZBCLJSY/ref=as_li_ss_tl?ie=UTF8&linkCode=sl1&tag=dcpsupplies-20&linkId=b61460804bc2873b0cd34aea912e1adf

There is not much I can say about a supply I haven't used.    The first thing that comes to mind looking at this supply is the quality of the binding posts which frankly don't inspire confidence based on that picture.   Which brings up a point with respect to binding posts, make sure these are covered:

• The post must be spaced on the industry standard.   That would be 3/4" or 19.05mm, at 19mm there is barely 2 thousands of an inch variance.
• Not only do you want the ability to accept banana plugs you want the posts to accept bare wires and spade type connections.
• Having a true earth ground reference point is very useful on a supply.   That post should be green or obviously metal connected to a grounded metal frame member.   What you DO NOT WANT is the ground connection shoved between positive and negative posts which screws up any potential use of ganged plugs.   When I look at that linked picture of the TechPower I immediately dismiss it as a meter for serious or even hobby use.
• The binding posts should be easy to replace with off the shelf parts when needed.

Now you may be thinking why in the hell is he going on about binding posts when everybody else is focused on ripple.    Well ripple can be important so I like to think I'm offering an alternative perspective here.   Binding posts are your mechanical interface to the power supply and the quality of those posts reflect immediately upon your opinion of the supply.  If you use a wire terminated with a spade connector and it keeps coming loose you will not be happy.   If a post strips out or otherwise gets damaged and you can't easily replace it you will not be happy.   If the banana post sockets are so loose you imagine that it lead a former life as a street walker you will not be happy.   If you take an off the shelf ganged banana plug connector and find that it will not fit you will not be happy.   Basically if a cheap binding post implementation is part of a supply, it is turned into a disposable piece of junk, rather that an instrument that can survive for decades on a bench.

By the way I fully support the idea of buying used that has been expressed here several times already.   For a beginner I also see huge value in what can be learned by DIY'ing a power supply (fixed voltage and simple to start).   Going used you might get a dud but it is also an opportunity to learn to repair.   In the end used hardware (in the USA anyways) is dirt cheap compared to new or even China imports.

[mindcrime]

Thanks again for all the advice. I went with the Rigol 832

Good call, IMO. I bought a DP-832A as my first "proper" bench power supply about two years ago and I've been very happy with it. The only significant "ding" against it is that the user interface isn't the most intuitive and convention thing in the world. But it's not so bad once you get used to it.