Insides of a 0-18V, 2A power supply, the CSI1802X

[slburris]

I recently purchased a CSI1802X from Ebay.  Normally I look
for supplies with LED displays, so I put in a low bid on it and
surprisingly picked it up for $26.

Details of the supply are here:

http://www.web-tronics.com/00addcreposu.html

Anyone have experience with Circuit Specialists?  Are they
a surplus dealer and/or OEM?  The supply has their name on
it, with a "Made In China" sticker.  Can't tell if it's their
design or a relabel of someone else (maybe Mastech)?

I ran some tests with my Fluke 87-V measuring voltage and
current.  The power supply displays were spot on, within
+- 1 digit of what the Fluke showed.

I'll attach pics of the inside in another post.

Scott

[slburris]

The insides seem surprisingly good.  Not sure if my expectations
are really low, or if this is on par for a supply retailing for $60.

The design is a straightforward linear supply, with a multitap
transformer and a relay that switches taps at certain voltage
points.

The display board is mostly surface mount.  No ECOs are visible.
There is something like silicone RTV on the power switch covering
any exposed metal space where the supply wires are attached.

The regulator board is phenolic, not FR4.  I guess it must save a
penny or two :-)

Scott

[saturation]

Wow, thanks for posting the detailed pics.  It looks like an older Mastech [ or Mastech copied them, who knows?] or maybe they all get their 'generic' boxes from the same maker and design the electronics around it.  Mastech lists the 1803 for $64 athough its +1A more than CS, the CS model appears decent, if it checks out spec wise.



Not mentioned in the manuals is that LED annunciator pictured above near the V adjust, its likely in constant voltage mode when it lights, is there another for constant current?


The newer Mastech's have a different look and adjusts, more akin to the GWInstek PSU:


I haven't had time to take photos, but compared to your CS PSU, the Mastech uses through hole components, no SMD, and very much like the photos I posted on this link:

https://www.eevblog.com/forum/index.php?topic=686.msg9661#msg9661

The links you provided include the schematics, what couldn't be better?  Its pretty straightforward too, the LCD readouts are built around the 7106 Intersil DVM, its dual, one for V and A. 

The PCB has QJE on them, and I can track it to a Chinese model PSU with model numbers QJXXYY for XX = volts, YY = Amps.

In China, the model referencing QJE directly are these, but from year 2008 :

http://www.gkzhan.com/st15602/product_304055.html
http://www.rclmeter.com/s02/product/2008/12/03/6443158.html

and you can see it has the same LCD board as the CS.

QJE direct Chinese links suggeset the parent model is made in the province Guangdong, where a QJE factory exists, but while they say they 'contract' manufacture their parent products are electrical equipment.

http://www.chinaqje.com/Company/

The PCB has clear hand soldering and corrections [see the link for the factory pics, lots of hand labor].

In the end, one thing we'd all like to know is how does it perform to spec?  Check the short circuit protection, you can also use that setting to check the constant current adjust.  Check the load regulation, just attached it to a load and put a scope across the power input and look for ripple. 

Once you've spec'd it out, that would be the 'real' specs I'd rely on rather than the rather spartan manual on the website.

Congratulations and enjoy!

[Kiriakos-GR]

slburris you need some training ASAP .. about resizing pictures... or just set your camera or mobile phone to 800x600 .

[saturation]

I like the pics just fine, its a good size until you click it, then you see the details!  It formats nicely until then.

Welcome back K, its good to read your always exciting  remarks 

slburris you need some training ASAP .. about resizing pictures... or just set your camera or mobile phone to 800x600 .

[Kiriakos-GR]

Well I am on true vacations ... 

And I do many things , and less walking over Internet.

The hot weather , and the peaceful  clean Blue sea , are all that I need right now. 

[slburris]

slburris you need some training ASAP .. about resizing pictures... or just set your camera or mobile phone to 800x600 .

I assure you the picture size is quite intentional.  If you don't need the detail, don't click on
the picture.  Each picture is labeled with the size and resolution, so you know what you're getting
into!  Especially for equipment teardowns, I like to be able to see the details.

I'm trying to upgrade the quality of the pictures I've been posting.  Most have been
taking with an original iPhone camera, i.e. they suck.  These were taken with a
Panasonic Lumix at high resolution. 

I'm finding that product photography is really hard to do well.  Someone posted
some stunning pictures in another thread, but I think they are using a DSLR
instead of a point and shoot like I am.

Scott

[saturation]

A great job.  Please keep posting BIG pictures, its the only way to see details enough like we are there, doing the dissection ourselves. 

slburris you need some training ASAP .. about resizing pictures... or just set your camera or mobile phone to 800x600 .

I assure you the picture size is quite intentional.  If you don't need the detail, don't click on
the picture.  Each picture is labeled with the size and resolution, so you know what you're getting
into!  Especially for equipment teardowns, I like to be able to see the details.

I'm trying to upgrade the quality of the pictures I've been posting.  Most have been
taking with an original iPhone camera, i.e. they suck.  These were taken with a
Panasonic Lumix at high resolution. 

I'm finding that product photography is really hard to do well.  Someone posted
some stunning pictures in another thread, but I think they are using a DSLR
instead of a point and shoot like I am.

Scott

[slburris]

Saturation: Yes, there's a constant current LED as well.

Great find on the OEM of this supply.  I sure wish there
was a geneology diagram of sorts of who make what for
who in China for power supplies (and test equipment)!

I did an overshoot test.  Set the voltage to 3.3v, fired
up the Rigol DS1052E (thanks Dave for making me buy it!)
and captured the turn-on.  It overshoots 4.2% to 3.44v.
Anyone know how typical this is?  Seems to settle down
in a couple of ms.  See attached photo.

Scott

[slburris]

How odd, you can buy the QJ2002A direct from China on Ebay
for $70 plus $55 shipping (220633831753) or get it from
Circuit Specialists in the US for $59 plus $10 for shipping.

That's not how it's supposed to work! 

Scott

[slburris]

I've connected a Luxeon K2 star as a load.  Set the voltage to about 7 volts,
then increased the current to about 800ma.  AC coupled to the Rigol and
I get about 44mv of noise with a frequency of about 416Hz, with a high
frequency noise superimposed.  See attached photo.

Scott

[alm]

That looks pretty horrible, since the spec is <3mArms in CC mode for I < 3A (pretty strange to specify ripple for I > 3A for a 2A power supply in my opinion), which shouldn't cause much voltage noise, since the I-V curve is exponential. Is it also this bad in constant voltage mode (over a resistor)? Sure it's not picked up by the wiring to the LED?

[slburris]

It *is* pretty horrible.  I was thinking maybe the supply was
hunting around when run in current limit mode, so I simply
set the voltage to 5 volts, didn't connect a load at all,
and remeasured.  The CV light was on instead of the CC.

Same result, low frequency 40ish mv ripple with a high frequency
overlay.

Just to be sure something wasn't spooking the Rigol DS1052E
I pulled out my Tektronix 2236 analog scope and looked at the
output.  Same thing.  I had it measure the high frequency
component, and it's an approximately 20mv sine wave with
a frequency a bit over 7Mhz.

Doesn't look like it meets its spec sheet.  However, I did buy
it second hand, not directly from Circuit Specialists, so I
don't know this supply's history.

As a power supply, it seems to function just fine, but it doesn't
meet its publish ripple specs.  Time to dig into the schematic and
do some debugging I think.

This unit isn't old enough to have dried out capacitors, but clearly
something's up.  Not sure I understand that 7Mhz oscillation either.

While I'm poking around at this, for you power supply experts out
there, if you were to make mods to this supply to improve it, what
would you do?

Scott

[saturation]

Yes, that's the simplest way to find out what any fundamental output problems are.  Specs are best at the max load.

The only thing to be wary off in Chinese products are substandard parts so they don't operate as specified, so the age of the unit may not matter.

The schematic has a signature that again mentions QJXXXXX and I translated the Chinese on the pdf to mean 'a Schematic Diagram of X series " via google.

The 40mV ripple can be suppressed by putting a large cap across the input, just rate it for the highest voltage the PSU outputs, say at 1000uF.  You can see the effect immediately on the scope.  If this works, then check if the output cap on the design if faulty and needs replacement.

The 400 Hz could be from the LCD driver circuit on the ICL7106 chip, but the typical oscillator frequency is 40kHz.

There are no other oscillators in this design.  I'm perplexed about the 7MHz pickup from somewhere.  This is high enough to be received as radiated radio frequency in the shortwave radio brand.  Is there a transmitter nearby?  It should be suppressed if the case is well grounded and the ground of your power is OK.  Interference could be riding on your AC line.  Check the output voltage from the secondaries of the transformer with your DMM, this insures its safe low voltage AC, then check the output wavefrom with your scope to see if its picking up the 7MHz.

Keep us posted!  The overshoot is not a problem but the ripple can be when you start using lower voltages in any design.[quote

BTW, the cheapo Mastech does output clean power at 3A, will post a review of its electrical qualities on a separate thread.

[author=slburris link=topic=973.msg12555#msg12555 date=1280720331]
It *is* pretty horrible.  I was thinking maybe the supply was
hunting around when run in current limit mode,so I simply
set the voltage to 5 volts, didn't connect a load at all,
and remeasured.  The CV light was on instead of the CC.

Same result, low frequency 40ish mv ripple with a high frequency
overlay.

Just to be sure something wasn't spooking the Rigol DS1052E
I pulled out my Tektronix 2236 analog scope and looked at the
output.  Same thing.  I had it measure the high frequency
component, and it's an approximately 20mv sine wave with
a frequency a bit over 7Mhz.

Doesn't look like it meets its spec sheet.  However, I did buy
it second hand, not directly from Circuit Specialists, so I
don't know this supply's history.

As a power supply, it seems to function just fine, but it doesn't
meet its publish ripple specs.  Time to dig into the schematic and
do some debugging I think.

This unit isn't old enough to have dried out capacitors, but clearly
something's up.  Not sure I understand that 7Mhz oscillation either.

While I'm poking around at this, for you power supply experts out
there, if you were to make mods to this supply to improve it, what
would you do?

Scott

[Simon]

maybe replacing the output capacitors and perhaps beefing up the values will help ?

[slburris]

So this is quite odd.  I put the scope on the base of Q1, and I see a 10Mhz low level oscillation there.
In fact, the Rigol reports is as being so close to 10Mhz that it's suspicious.  The Tek 2236 reported a
7Mhzish signal at the output, but I'm not sure how much faith I have in it's measurement, especially
since it's a high frequency sine wave imposed on a lower frequency sine wave.

In any case, having that oscillation at Q1 isn't a good thing as it will get propagated through
to the output via Q2.  If you look at the photos, that base signal is carried by that yellow wire
that stretches from the front to the back of the case.  That seems like a poor design choice.
I would have made that a shielded wire myself.

The more I stare at the schematic, the more I'm finding that it's only approximately correct.
There's a 470uf cap across the output on a little PCB which contains the output jacks and I
can't find that anywhere on the schematic.

I'm really puzzled by the 10Mhz signal though.  There is a 10Mhz house standard running
to several instruments, but I wouldn't have thought the power supply would be able to
pick that up.  Maybe I need to carry the supply to some other part of the house and
see if that goes away.  Still no idea about the 200Hzish lower frequency ripple yet.

Scott

[slburris]

I've studied the hi-res photo of the display board, and I take back
my statement about the quality being surprisingly good.  Either this
board had extensive rework, or this board was hand soldered.
Heck, it looks like the boards *I* would hand solder :-)

Look at the 0 ohm jumper DP3, R59, C10, R50, R23, R22.
All show signs of sloppy work, either poorly aligned, too much solder,
or both.  Lots of solder flux around these areas too.

And look at the yellow wire that goes to the regulator board.
It's just tacked to one of the pads of D2!  That's absurd, there's
a PCB pattern for CN2, presumably a connector of some sort,
and the wire could have been soldered in one of the CN2 holes at
least!

[saturation]

I find this weirdness fascinating.

I'd need to first insure the MHz signal isn't coming from the power lines, and the easiest way is to check it on the secondaries of the transformer, before any electronics.  If there isn't any there, then for whatever reason its being generated by the components in the PSU.  It can happen from parasitic oscillations, but I haven't seen those in designs without a clock somewhere in the circuit, and its too remotely connected from the digital LCD meter section which is clocked at a much lower frequency.

The PSU schematic does not have any filters for MHz frequencies, no bunches of coils, and ferrite cores and beads in there if it did, like in switching supplies.

What do does the last highlighted statement mean, "10Mhz house standard running to several instruments", do you have a home automation network running through the power lines?  If so, that can trigger parasitic oscillations in some transistors.

So this is quite odd.  I put the scope on the base of Q1, and I see a 10Mhz low level oscillation there.
In fact, the Rigol reports is as being so close to 10Mhz that it's suspicious.  The Tek 2236 reported a
7Mhzish signal at the output, but I'm not sure how much faith I have in it's measurement, especially
since it's a high frequency sine wave imposed on a lower frequency sine wave.

In any case, having that oscillation at Q1 isn't a good thing as it will get propagated through
to the output via Q2.  If you look at the photos, that base signal is carried by that yellow wire
that stretches from the front to the back of the case.  That seems like a poor design choice.
I would have made that a shielded wire myself.

The more I stare at the schematic, the more I'm finding that it's only approximately correct.
There's a 470uf cap across the output on a little PCB which contains the output jacks and I
can't find that anywhere on the schematic.

I'm really puzzled by the 10Mhz signal though.  There is a 10Mhz house standard running
to several instruments, but I wouldn't have thought the power supply would be able to
pick that up.  Maybe I need to carry the supply to some other part of the house and
see if that goes away.  Still no idea about the 200Hzish lower frequency ripple yet.

Scott

[saturation]

Yes, that's why you have great photos! They show details.  It is sloppy, there are also areas of solder in the wrong places, but in the end the proof is in units capacity to deliver its 18V at 2A to its spec.  It has past the turn on test, and does provide output albeit with ripple and interference, and I presume, it still getting more evaluation.

I've studied the hi-res photo of the display board, and I take back
my statement about the quality being surprisingly good.  Either this
board had extensive rework, or this board was hand soldered.
Heck, it looks like the boards *I* would hand solder :-)

Look at the 0 ohm jumper DP3, R59, C10, R50, R23, R22.
All show signs of sloppy work, either poorly aligned, too much solder,
or both.  Lots of solder flux around these areas too.

And look at the yellow wire that goes to the regulator board.
It's just tacked to one of the pads of D2!  That's absurd, there's
a PCB pattern for CN2, presumably a connector of some sort,
and the wire could have been soldered in one of the CN2 holes at
least!

[alm]

I would be careful with permanently increasing the capacitor values, especially at the outputs. It is obviously a fine test. This capacitor is discharged without current limiting when you short the outputs, so you don't want a huge capacitor that can be used for spot welding here. The SMT soldering looked pretty sloppy to me, too.

[slburris]

What do does the last highlighted statement mean, "10Mhz house standard running to several instruments", do you have a home automation network running through the power lines?  If so, that can trigger parasitic oscillations in some transistors.

I have a Trimble Thunderbolt providing a GPS locked 10Mhz frequency standard feeding to
a TAPR TADD-1 Distribution Amplifier (http://www.tapr.org/kits_tadd-1.html)

This is fed to several frequency counters and signal generators to their external oscillator
inputs.  All done with coax cable, so I wouldn't expect the power supply to pick it up, but who knows?

Scott

[saturation]

Thanks, I think this is very related to the problem, its leaking from somewhere in the distribution or source.  Its very unlikely the frequency you've chosen and the interference in the PSU are coincidentally 10MHz.  The 7 MHz you mentioned earlier could be this, just that its not triggering properly on the scope.

What do does the last highlighted statement mean, "10Mhz house standard running to several instruments", do you have a home automation network running through the power lines?  If so, that can trigger parasitic oscillations in some transistors.

I have a Trimble Thunderbolt providing a GPS locked 10Mhz frequency standard feeding to
a TAPR TADD-1 Distribution Amplifier (http://www.tapr.org/kits_tadd-1.html)

This is fed to several frequency counters and signal generators to their external oscillator
inputs.  All done with coax cable, so I wouldn't expect the power supply to pick it up, but who knows?

Scott

[slburris]

Bah!  Double Bah!  Triple Bah!

I powered down the Trimble Thunderbolt and the high frequency signal went away
and the power supply appears to now be well within it's noise specifications.

What lead me astray was the initial measurement on 7Mhzish on the Tektronix 2236.
Maybe the signal was noisy enough that the frequency counter in the 2236 was getting
confused.  The Rigol DS1052E that I later used showed it at 10Mhz.

So the power supply is good, if a bit sloppily assembled.  And I've been apparently
leaking 10Mhz RF for some reason (maybe a bad cable?).  Huh, I wonder if that's
affected reception of WWV on my Heathkit GC-1000 clock in the next room!

Scott

[alm]

Did turning of that Thunderbolt also get rid of the ~400Hz noise?

I wonder if inadequate shielding/filtering in the power supply contributed to this issue, although it may have been picked up by your test leads, or because you took the case off.

[slburris]

Did turning of that Thunderbolt also get rid of the ~400Hz noise?

Yes it did.  Some sort of beat against the 10Mhz somehow?

I would have thought that the power supply would have been shielded
enough when fully assembled, but I'm no RF expert.

Now I have no complaints about my $26 power supply :-)

Scott