Keysight New instruments

[bdeianov]

Or just stuff a diode in series. 

That's what the manual tells you to do.
Unfortunately no external sense to compensate for the drop.

I was curious what the E36312A / E36313A supplies do, which do have remote sense. Turns out they have the same note in the manual. They further show the sense line connected before the diode, so they still can't compensate for the diode drop.

[Berni]

Same thing is recommended for the rack mount system PSUs 6621A, 6622A,6623A, 6624A,6627A



But i have feed voltage back into one of these supplies and it behaved nicely. If you try to pull the output voltage above what it is set to it simply starts sinking current to keep it in regulation. I have even used it as a "battery simulator" on a power bank and it was perfectly happy both providing current and sinking current. The sinking ability appeared to be limited by the channels maximum rating. So if you had a 60V 1A channel it would sink up to 1A of current before the voltage goes out of regulation. All this is done in analog circuitry and works well. So yes it does sink current but it would just slowly discharge the battery.

However the manual there talks about OVP specifically. This is a feature that is off by default and you have to manually turn on using a button. I have not tested that, but i would assume it does the usual sinking up to the channel maximum, but if you shove in more current that the channel can handle then the voltage starts to rise, eventually hitting the (likely in hardware) OVP trip point. At this point it would make sense for the PSU to fire off a SCR across the output to short it out in a last ditch attempt to keep the voltage from rising too high.

Most other PSUs have an internal pre-load resistor across the output that is behind the current sense shunt. This resistor mostly being there to discharge capacitors and help the linear regulator be more stable. So as a result most PSUs do have sinking capability but only in the 10s of miliamps (depends on the voltage)

[2N3055]

Problem here is specifically in crowbar circuit. PSU that are more than one quadrant, or have active down programmer have current limit and won't be damaged. Most of the PSU are specced for parallel mode, and if they are, they are designed to back fed the voltage.  Problem is only if crowbar circuit fires, or you reverse voltage, and that only if outside source (battery) can source too much current ( which decent sized SLA or LiIon or LiPO can) so there is damage.

Bottom line, we all on occasion charge batteries on those lab PSU, but they are not meant to do so...
If I do, I put big Schottky diode in series and raise voltage 0.5V. That's close enough for charging.
Or I could put in mosfet protection reverse voltage circuit  and not worry about voltage drop at all. That one could be put in a small box with cables on and you just grab charging cables and connect battery...
It's just I don't do this often enough  to be bothered to make one...

[tautech]

If a lab PSU isn't safe to charge any battery it's not worth the scrap metal it's made from ! 

At power loss it should default to output OFF unless it has an option to override this.
OVP need also be selectable and enabled as necessary where instead of blowing fuses it should trigger the output to OFF.

None is magic and are standard features in some PSU's.

[Bud]

Those rack mountable HP power supplies "can" sink current but that was not meant to serve as electronic load, and their operator's manuals say that. Current sink feature was meant for programmable testing, such as evaluating Load's regulation and transient response. A program would run via GPIB, making the power supply change voltage fast. When such changes go from higher voltage to lower voltage , the power supply has to be able to sink current from charged capacitors on the Load side.

[bd139]

Or just stuff a diode in series. 

That's what the manual tells you to do.
Unfortunately no external sense to compensate for the drop.

I was curious what the E36312A / E36313A supplies do, which do have remote sense. Turns out they have the same note in the manual. They further show the sense line connected before the diode, so they still can't compensate for the diode drop.

What about using a MOSFET as an ideal diode circuit instead of just a diode?

[2N3055]

Or just stuff a diode in series. 

That's what the manual tells you to do.
Unfortunately no external sense to compensate for the drop.

I was curious what the E36312A / E36313A supplies do, which do have remote sense. Turns out they have the same note in the manual. They further show the sense line connected before the diode, so they still can't compensate for the diode drop.

What about using a MOSFET as an ideal diode circuit instead of just a diode?

See my post above

[bd139]

That'll teach me for not reading all the posts 

[Fungus]

Some supplies have a fuse...

My $35 FNIRSI even came with with two spares in SMD tape: 

Or just stuff a diode in series. 

That's what the manual tells you to do.

Let's hope the students remember to do that. 

[nctnico]

If a lab PSU isn't safe to charge any battery it's not worth the scrap metal it's made from ! 

Nonsense. For charging batteries you use a battery charger which is suitable for the chemistry. A lab PSU is the wrong tool for battery charging so don't be surprised it is not 100% suitable for that purpose. The function of overvoltage protection / crow bar is to protect your potentially expensive DUT (a prototype SoC board can easely cost US $1000) against operator error or defects inside the PSU.

[Fungus]

If a lab PSU isn't safe to charge any battery it's not worth the scrap metal it's made from ! 

Nonsense. For charging batteries you use a battery charger which is suitable for the chemistry. A lab PSU is the wrong tool for battery charging so don't be surprised it is not 100% suitable for that purpose. The function of overvoltage protection / crow bar is to protect your potentially expensive DUT (a prototype SoC board can easely cost US $1000) against operator error or defects inside the PSU.

Battery protection modules are about $0.15 on Aliexpress: https://www.aliexpress.com/wholesale?SearchText=TP4056

[SilverSolder]

[...] A lab PSU is the wrong tool for battery charging [...]

Is that totally fair?  -  a lab PSU is a general purpose source of power that shouldn't really care what you use it for (within reason!).  A battery charger is an application specific device.  So, I would accept that the charger is "better"...  but surely I should feel OK about connecting anything I like to my lab power supply that isn't directly insane? 

For example, the other day I was measuring the internal resistance of a bunch of NiMH cells.   My battery charger was not able to do this...  so I used my ancient lab power supply in constant current mode and measured the difference in voltage drop between two currents, differing by 200mA.   So, connecting a battery to the supply is a possible use case even if you are not charging it...   

Also, consider the use case of charging a supercap, or even a regular capacitor (massive capacitor bank in an audio amplifier, for example) and subsequently decide to lower the voltage...

Overall, it seems preferable to me that a lab supply is able to deal with these kinds of real world scenarios without the risk of self destruction.  Granted, a crowbar is always going to be an issue and probably should not be used without a fuse...  but that could also be baked in.

[bdunham7]

Nonsense. For charging batteries you use a battery charger which is suitable for the chemistry. A lab PSU is the wrong tool for battery charging so don't be surprised it is not 100% suitable for that purpose. The function of overvoltage protection / crow bar is to protect your potentially expensive DUT (a prototype SoC board can easely cost US $1000) against operator error or defects inside the PSU.

The manual for mine specifically lists battery charging as a function and gives details on how to do it.  A 'correct' charger is optimal for routine battery charging, but people do test/repair/characterize/analyze/etc battery systems just like anything else and there are many times I'm interested in more than just charging the battery.  Other times I may need to charge an internal or odd battery in order to test something and the 'correct' charger is broken or unavailable.  It's an everyday use for me and I can't think of anything that would work better.

 It doesn't have a separate OVP, but designing an OVP should be straightforward and properly implemented it shouldn't result in damage if you connect a battery.  In non-battery situations where the OVP is there to protect your SoC from programming errors or internal control circuit failures, triggering the OVP should result in a current-limited internal short or, if the current-limiting part fails, a blown internal fuse.  Triggering the OVP while a battery is connected should result in a blown fuse.  Whether or not the PSU is capable of sinking current (mine does not) shouldn't matter much to a good OVP design.  If you don't need or want battery charging capability then it is fine to buy a PSU that 'fails' the battery test, but there isn't any inherent conflict that I can see between the two goals of protecting your SoC and not going up in flames if someone connects a battery.

[nctnico]

Now try to figure out how HP never came up with the idea of having a fuse in series with the output of a PSU

[exe]

Disagree entirely. Crowbar is important when the following equation is true:

cost_of_load > cost_of_power_supply

A year or so ago I had what i consider to be a fairly decent quality supply (HP) blow the reference zener. This caused the output to bang itself at the collector supply voltage. You should plan for such failures.

How does a crowbar work? I assume the expensive load is also sensitive to overvoltages. In this case the crowbar should engage very fast and at voltages not so much above the set voltage. If reference circuit doesn't work,  how does the crowbar know when to trip?

If crowbar set to trip at, say, 30V, then it's already late, imho.

[SilverSolder]

Now try to figure out how HP never came up with the idea of having a fuse in series with the output of a PSU

I guess you'd have to sense the voltage on both sides of the fuse to do a really good job...  it starts to get expensive! 

[HighVoltage]

The problem I have with crowbar is that I did not know about it, until I read the manual.

The brand new Agilent PSU that I killed a few years back, was a E3648A series PSU.
The heavy crowbar traces were burned away completely.
Luckily the traces could be repaired with heavy copper wires.
Now I am aware !


 

[bdunham7]

Now try to figure out how HP never came up with the idea of having a fuse in series with the output of a PSU

You tell me. Some other manufacturers have them, some don't.  I've worked on quite a few bench PSUs in recent years and although later model HPAK PSUs like the E36xx series are expensive and widely held in high regard, they don't impress me all that much.  There's no reason to let them go up in flames rather than simply have a large fuse to protect against gross overloads.  Or simply use an OVP system other than a crowbar.

[Berni]

I think a fuse would be a good idea to have on a PSU with such a SCR crowbar circuit.

Its not even that difficult to get around the fuse voltage drop. All you do is hook up the sense wire after the fuse. Since the voltage sense is a high impedance input it is a lot easier to protect it using the good ol resistor and TVS diode.

Tho that fuse then only protects the PSU from the DUT. If the crowbar trips because of a fault inside the PSU it wont blow, but then again at that point you need an extensive PSU repair anyway, and a common fault with linear PSUs that makes them output too much voltage is a blown pass transistor, so your power stage is already dead anyway, eventually a fuse at the transformer would blow to stop it from catching fire.

[tautech]

Why is it that foldback current limiting isn't incorporated in a 'quality' PSU FFS, even a $30 SMPS has such a feature as it's hardly a new leading edge technology when it's been around for decades.

[bd139]

Disagree entirely. Crowbar is important when the following equation is true:

cost_of_load > cost_of_power_supply

A year or so ago I had what i consider to be a fairly decent quality supply (HP) blow the reference zener. This caused the output to bang itself at the collector supply voltage. You should plan for such failures.

How does a crowbar work? I assume the expensive load is also sensitive to overvoltages. In this case the crowbar should engage very fast and at voltages not so much above the set voltage. If reference circuit doesn't work,  how does the crowbar know when to trip?

If crowbar set to trip at, say, 30V, then it's already late, imho.

Usually, taking radio power supplies as an example, you set the voltage to 13.8V and the crowbar to 15V. The specification for the load sets the voltage tolerance. Typically this is max input voltage of LDOs, capacitors, ICs etc directly connected to the supply considering power dissipation as well. Some things designed for say 9v input will quite happily run at 12v depending on how the internal power is arranged.

Crowbar usually has a voltage ref and comparator and SCR trigger and that’s about it. They are extremely fast.

There are two cases where the crowbar comes in handy as well which are almost never mentioned as well. Firstly, and this does happen with RF stuff, when there is RF noise on the sense lines to the supply. Secondly, when the load is too inductive and the supply starts oscillating. Even the best of power supplies can be coaxed into oscillating (try running something like an SMPS on the end of a few metres of untwisted wire.

Why is it that foldback current limiting isn't incorporated in a 'quality' PSU FFS, even a $30 SMPS has such a feature as it's hardly a new leading edge technology when it's been around for decades.

HP6236/6237 have fold back limiters on the third supply. This is only used above about 500mA where the load is likely to blow up in your face or turn incendiary. I think it’s more a “let’s not burn the place down until the next engineer does a walk past of this test rig” feature b

[tautech]

Why is it that foldback current limiting isn't incorporated in a 'quality' PSU FFS, even a $30 SMPS has such a feature as it's hardly a new leading edge technology when it's been around for decades.

HP6236/6237 have fold back limiters on the third supply. This is only used above about 500mA where the load is likely to blow up in your face or turn incendiary.

Wow, on the 3rd supply.....well I guess that's something. 
It might seem the designers couldn't count when the only are 2 more outputs.   

[bd139]

Well the other two are only 500mA so 

Also it's difficult to rationalise foldbacks when you have a variable current limit. I mean you can't go over-current. What if it's a CC source you want (charging SLAs). Foldback would be a problem.

[nctnico]

Why is it that foldback current limiting isn't incorporated in a 'quality' PSU FFS, even a $30 SMPS has such a feature as it's hardly a new leading edge technology when it's been around for decades.

How would that be compatible with constant current mode? How is a outputing a random voltage at a random current going to help?  AFAIK foldback current limiting is used to limit the dissipation inside the PSU in case the heatsink is not big enough.

Besides that SMPS don't have fold-back current limiting; they just go into 'hickup mode' sending pulses into the load.

[Berni]

Well the HP 6032A does have a dedicated foldback button.



It works like a normal bench PSU normally, but if you push the foldback button the LED next to it lights up and this locks the PSU into the current mode of operation. If you are currently in CV mode and the supply hits CC mode then it will instead shut down and wait for the output to be disabled and enabled again. If you are in CC mode (say powering a big LED with a set current), push the foldback button and then disconnect the load, causing the PSU to go back into CV mode it will shut down too.

Such a feature is probably even more desirable on such a massive PSU that can supply 1kW 60V 50A.