EEVblog #828 - Siglent SPD3303X precision Lab PSU Teardown

[John Coloccia]

Crap binding posts and non-standard spacing is all I needed to see. Geez, here I am watching this video because I'm in the market for a new power supply and I'm willing to give Siglent a chance to rehab itself, and it all goes to crap in the first 3 minutes. WTF? 

[tautech]

Does the display really turn off after some "inactivity"? Even if it allows you to turn that off, they made it the default?

As Trio supplied this PSU, who know if it was fiddled with before Dave got his hands on it, as these have just been released it was probably their demo unit.
Did Dave look at the menus for screen saver....no.
Did Dave look for a default menu......no.

Let's not judge to hard until he does a comparison and spends some time in the UI.

As for the binding posts, I agree. 

[FrankBuss]

What's the current capacity of that supply? Like I said, I don't like a supply in 2015 that doesn't have a current regulated output. Unfamiliar caps I can live with if the quality is there and I think Dave should be less harsh on his condemnations.

Basically it's a dual supply, not a triple, the DP832 is a real triple whatever other warts it might have.

The SPD3303D and SPD3303X have nearly the same current capacity, see here and here. Both are triple output power supplies. If you don't need 1 mA / 1 mV resolution (doesn't make much sense without extra sense inputs anyway for higher currents), I think the price of the SPD3303D is very good for the product, if you can live with the minor disadvantages, like that you can adjust the voltage in 1 V or 10 mV steps only, not 100 mV, and yes, the screen-saver is a default on the SPD3303D, I didn't find a way to disable it. The GUI and the construction inside looks very similar to the SPD3303X, so I guess there is no way to disable it on the SPD3303X either. Looks like the other main difference is that the SPD3303D has no ethernet. But I've tested it and it works without problems over USB (at minute 5:00 in the video).

[Hydrawerk]

What is the difference between SPD3303D and SPD3303X?
https://www.eevblog.com/forum/testgear/siglent-spd3303d-review/

[Lightages]

Dave, I think you went a bit easy on that power supply. The bad construction techniques to on the sharp edges rubbing against the mains wire, bad binding posts and non-standard spacing, and the ultra high output capacity.... all for over $500? Sorry, its a fail.

[Huluvu]

they use PWBs without vias ....



I cannot believe this to see it at a 600 Euro PSU.

The Lelon Brand Caps are typical for Buyers keeping the BOM Cost down. 

[FrankBuss]

What is the difference between SPD3303D and SPD3303X?
https://www.eevblog.com/forum/testgear/siglent-spd3303d-review/

See the webpages:
http://www.siglent.eu/siglent-spd3303d.html
http://www.siglent.eu/siglent-spd3303x-power-supply.html
If I didn't miss anything:

SPD3303D	SPD3303X
3A	3.2A
0-30V	0-32V
10mV/10mA resolution	1mV/1mA resolution
USB	USB and ethernet

I wouldn't pay the 209 EUR more for the SPD3303X. But for some applications the higher resolution might be useful, even if you don't get the accuracy at the end of the cable.

[Salas]

Interestingly all my banana terminated chords fit well in the output posts of my SPD3303S (previous model sans network connector). Not too loose not too tight even if the banana ends are flared or ridged. Weird, because they are the same type 4mm binding posts.

[Salas]

Does the display really turn off after some "inactivity"? Even if it allows you to turn that off, they made it the default?

I recently bought some cables (4mm banana) listed as "unbranded" but very cheap. I can't use them, but I guess they'd fit this 539USD power supply.

--
Mihai

Yes, after half an hour of not pushing buttons or turning the knob.

[bigsky]

Dave, at 14:15, the components you say are PTCs are actually polyfuses - I could tell from the shape and googling the number confirms this - see www.wondhope.com/pptc/WH250.html

[Arjan Emm]

If you can't properly connect a load to a powersupply, well, nothing else matters really. End of story. Big fail on the binding posts! Accuracy 1mA and 1mV, rather useless if it changes by 50 if you touch the banana plug. Esp. without sense connection.

[RupertGo]

I really don't like that slide switch on the third output. Very 70s. They've got a big ingress area for dirt and what have you, and if you brush against it while you're moving your hand around the front panel, it could easily shove 5v down your 2.5v logic.  Why not a cycling push button or, even better, a proper twisty knob? Or even control it from the main input knob? I bet it can't be controlled remotely either (not that you'd normally want to).

That feature looks like an afterthought.

[Fungus]

I really don't like that slide switch on the third output. Very 70s. They've got a big ingress area for dirt and what have you, and if you brush against it while you're moving your hand around the front panel, it could easily shove 5v down your 2.5v logic.  Why not a cycling push button or, even better, a proper twisty knob? Or even control it from the main input knob? I bet it can't be controlled remotely either (not that you'd normally want to).

I don't see how a button or knob is less likely to move accidentally (a knob is probably easier).

PS: I agree with everybody else; those output posts are an instant deal breaker. How did they even get into a production PSU? Did they get a truckload of reject connectors for free...?

[blueskull]

Why there are still people making linear power supplies besides these really high end SMUs?

Why not SMPS+post regulator architecture? Less power dissipation, and comparable noise and ripple.

Anyway from a different chassis there is no way to get super clean power source, so for serious analog circuits, there will always be local regulators.

A comparable resolution can be achieved from a BK9110, which is a little bit more expensive per channel, but definitely much more powerful.

For me, I don't use 2.5V/3.3V/5V from PSU, I usually get them from USB, so I only need a 2 ch PSU. In this case, BK9110 is actually cheaper than SPD3kX per channel.

[TheSteve]

Really seems overpriced for the quality, another Siglent product to avoid.

[c4757p]

Why there are still people making linear power supplies besides these really high end SMUs?

Why not SMPS+post regulator architecture? ... comparable noise and ripple.

Comparable? To an all-linear? I'd like to see you pull that off.

Who gives a damn about efficiency in a lab supply?

[blueskull]

Why there are still people making linear power supplies besides these really high end SMUs?

Why not SMPS+post regulator architecture? ... comparable noise and ripple.

Comparable? To an all-linear? I'd like to see you pull that off.

Who gives a damn about efficiency in a lab supply?

9110 HAS a linear regulator, follows a DC/DC, which is fed from an offline SMPS.

Their datasheet stats 2mV ripple at worst case, while the worst case ripple for a GW Instek GPS series is 1mV. The new Siglent one stats 1mV too. Both RMS values.

[rdl]

Maybe hard to describe, but I didn't like the single black digit and background highlight on the display when adjusting and it looked like it never went away. If not being adjusted all numbers need to look the same.

I also agree 100% as to the following physical issues, which are no buy faults in my opinion.

1 Rusty sharp metal edge against mains wiring (fixable)
2 Nearly useless binding posts (replaceable)
3 Capacitors which can't be trusted (replaceable)
3 Unsupported TO-220 devices (no easy fix, needs new pcb layout)

I may have missed other issues as I lost interest around the 10 minute mark, when it became obvious this was over-priced junk.

It's unbelievable to me that they would still be shipping product with factory installed rust. Because of that alone, I doubt I will waste another second looking at, or considering a Siglent product of any kind.

[Salas]

Why there are still people making linear power supplies besides these really high end SMUs?

Why not SMPS+post regulator architecture? ... comparable noise and ripple.

Comparable? To an all-linear? I'd like to see you pull that off.

Who gives a damn about efficiency in a lab supply?

9110 HAS a linear regulator, follows a DC/DC, which is fed from an offline SMPS.

There datasheet stats 2mV ripple at worst case, while the worst case ripple for a GW Instek GPS series is 1mV. The new Siglent one stats 1mV too. Both RMS values.

Reminds of KORAD looks. Got to confirm on a scope it really filters out SMPS patterns across a wide band. Also find out if interference is escaping towards the AC chord or it propagates out of the box. For general use almost anything is OK until you FFT a low signal high gain circuit for instance. Rail dirt changes its noise floor grass into a jungle.

[AF6LJ]

Nice $350.00US power supply.

[c4757p]

Why there are still people making linear power supplies besides these really high end SMUs?

Why not SMPS+post regulator architecture? ... comparable noise and ripple.

Comparable? To an all-linear? I'd like to see you pull that off.

Who gives a damn about efficiency in a lab supply?

9110 HAS a linear regulator, follows a DC/DC, which is fed from an offline SMPS.

Their datasheet stats 2mV ripple at worst case, while the worst case ripple for a GW Instek GPS series is 1mV. The new Siglent one stats 1mV too. Both RMS values.

If these things have ripple that is that bad, then linear power supplies have got way crappier since the nice 60s-80s PDI ones I'm used to.

Nice $350.00US power supply.

Nice? You'd spend $350 to get some shoddy banana jacks that don't hold plugs and some mains wiring that's going to short to the chassis in a few years?

Nah. This PSU put Siglent back on my shit list, actually. Releasing a product with mains wiring like that is practically malicious.

[ErikTheNorwegian]

The binding posts are a killer turnoff..


Dave, what about a test of this powersupplay.. killer price, done in 15 minuts to rewire it to 220V.. very popular. Its the same as HANTEK.

http://www.ebay.com/itm/221885966366?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Circuit Specialists PPS2320A Tri Output Linear Programmable Bench Power Supply

Key Features:

    3 Channels (2 each 0-32 V/3A) and 1 fixed multiple output 2.5V/3.3V/5V @ 0-3A
    64V Max Output Voltage (32 V per Chanel)
    6A Max Output Current (3 A per Channel)
    Modes:Constant Current/Constant Voltage/SW Control/Parallel(double Current), Serial(double voltage). Split-Rail (Negative and Positive Voltage)

This 3 Channel Benchtop Power Supply is a mult-functional programmable power supply featuring 2 variable outputs (0-32V/0-3Amps) and 1 fixed output channel that the user may select either 2.5V, 3.3V or 5V from 0-3 Amps. This lab grade product boasts all digital controls from the front panel or may be controlled via PC interface.

[blueskull]

If these things have ripple that is that bad, then linear power supplies have got way crappier since the nice 60s-80s PDI ones I'm used to.

Yes, lab power supplies suck. Anyway ground noise will render even a perfect PSU noisy if you consider sub mV "noisy".

That's why there are usually LDOs on board, unless it is a digital board without very high frequency PLLs.

[Macbeth]

I seem to recall there was a batch of Rigol DP832's with out of spec binding posts - they were on the smaller gauge so nobody could get their 4mm banana plugs to fit.

Is there an actual international standard for these binding posts and plugs? If they don't meet it send it back.

Having said that I was surprised at my recent acquisition of an ancient ESI Dekavider. The terminals are lovely, all tellurium copper, but they seem to be around 4.25mm ID and are only 0.5" deep so a banana plug sits loose and and wobbly. Just what was the standard back in the 60's?

[langwadt]

- Deceptive Corporate practices
 - Non standard width banana plugs
 - Crap quality banana plugs
 - Shit caps

And the shorted turn of the toroidal transformer. The mounting screw doesn't seem to be isolated. Rust, again. The bodged ethernet. Build quality is the same like my Mastech clone. I wouldn't pay more than €300 for that Lab PSU.

if the turn was really shorted you would know, a 300W transformer with a ~0.5V turn shorted, something would be glowing