--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/content/HPZR24W.rst	Fri Sep 22 00:13:22 2017 +0200
@@ -0,0 +1,195 @@
+=====================
+ HP ZR24W PSU Repair
+=====================
+
+:Author: David Douard
+:Category: Electronics
+:Tags: ZR24W, PSU, PWB-1336-02, TPS-10036
+
+A friend of mine had a problem with his (a bit old) 24" HP monitor: the device
+was working properly, but only for a while (few minutes at most).
+
+Looked like a good candidate for a quick repair, problem being most probably
+bad caps are so.
+
+The first step was to find out how to disassemble the down thing, since these
+"modern" things are not designed to be serviced and have no screw, only plastic
+clips which are so easy to break.
+
+Whatever, I finally succeded in opening the box (with only one or 2 broken
+clips).
+
+Overview
+========
+
+Inside the beast, the usual stuff under the shields:
+
+- a PSU,
+- a main controller board,
+- an LCD driver board, and the monitor being a bit old,
+- a blacklight HV driver board.
+
+
+.. image:: {filename}images/ZR24W/back.jpg
+   :alt: Back of the HP ZR24W
+
+First thing, to make sure the problem comes from the PSU, I powered the monitor
+from my bench power supplies. I used one output of my HP E3648A to generate the
+19V rail that powers the backlight. It was a bit short (CC limited at 16V), but
+enough to power the backlight. The second output was used to produce the 12V,
+and the 5V was generated by my very old home made PSU.
+
+.. image:: {filename}images/ZR24W/ext_power.jpg
+   :alt: Testing the display with my bench PSU.
+
+The monitor was working fine when powered from these PSUs, so the problem was,
+indeed, this Tatung PWB-1336-02 switching PSU:
+
+.. image:: {filename}images/ZR24W/psu_top.jpg
+   :alt: Top of the Tatung PWB-1336-02 PSU
+
+.. image:: {filename}images/ZR24W/psu_bottom.jpg
+   :alt: Bottom view of the Tatung PWB-1336-02 PSU
+
+The design is nice and clean. Electrolytic caps however are not the best ones
+(mostly Lelong ones I think), which is not really a surprise in this kind of
+device.
+
+At first sight, nothing strike the eye, no leaky cap, no burnt resistor or
+PCB...
+
+Not a 10mn fix, in the end.
+
+The PSU generates 3 voltages:
+
+- 5V  @2.7A
+- 12V @0.8A
+- 19V @3.2A
+
+
+Finding the problematic power rail
+==================================
+
+There is a hard switch on the PSU, next to the IEC socket, and When the power
+is on, the 5V is hot, wether the monitor is on or in standby.
+
+When the display is powered on, the 2 other voltages are started.
+
+The pin 10 on the cable between the PSU and the main CPU board is dedicated
+to the "power saving" state. It must be high (at 5V, which is always present)
+to enable the 2 other voltages.
+
+Note that there are 2 other pins dedicated to power management (pin 11 and 12,
+marked as "On/Off" and "Vadj"). But these are directly routed to the backlight
+board and take no part in the PSU management.
+
+
+My first test has been to plug my cheap `electronic load
+<{filename}/ZPB30A1.rst>`_ on the 5V with the 2 other voltages stopped.
+
+And I could reliably sink 3A from there. So the problem must be on one of the 2
+other rails.
+
+But I also could sink the max amperage from the 2 other power rails (1A from
+the 12V and 3.5A from the 19V)...
+
+Ok, so each power rail seems to work fine alone.
+
+But when I sink current from the 5V rail while the 2 other voltages are up,
+then the PSU fails after a short while.
+
+I've tried to probe a bit the switching curves using my Rigol DS1054, but the
+PSU being "hot" (around 400V), and having no isolation transformer, I could not
+probe the signals correctly (using 2 probe and displaying the A-B curve, which
+is some kind of a joke on the Rigol, since you cannot hide the A and B curves:
+the curve substraction is purely computed from displayed curve. When I found
+how useless this was on the Rigol, I had not enough energy to extract my old
+`Tek 2445A <{filename}/TeK2445.rst>`_ from under the pile of test equipment
+stowed in the closet...)
+
+
+A glimpse at schematics
+-----------------------
+
+At this point, I needed to try to understand a bit the schematic. I wasn't even
+sure the problem came from this side of the PSU, it could also be a problem in
+the "isolated" part (bad caps, bad optocoupler, bad voltage reference...). I
+had checked the main caps, and they seemed to be ok (not quality japanese
+brands, but still the correct value and low ESR).
+
+
+On the live part, before the transformers, there are 2 chips :
+
+- a TNY279PN_ dedicated to the 5V rail
+- a CM6807_ for the 19V rail
+
+The 12V is produced from the 19V rail by a small DC to DC converter (FP6185_).
+
+In fact, this PSU design is almost just the 2 application circuits (found in
+their respective datasheets) merged together.
+
+The noticable points are:
+
+- The TNY279PN DC input can be come from either the input bridge rectifiers
+  (via a diode and a thermistor), or, when powered, by the DC produced by the
+  CM6807 via the PFC circuit. So when the CM6807 is down (when the monitor is
+  in standby), the DC input for the TNY279 is around 318V, but when the monitor
+  is on, the CM6807's PFC circuit rise this voltage to almost 400V.
+
+- The bias winding, normally used to power the TNY279 (to allow a very low
+  no-load power consumption, datasheet says <50mw) also powers the CM6807.
+
+  A funny side effect of this design is that the CM6807 cannot be started if
+  there is no load on the 5V rail: in this case, the auxiliary voltage of the
+  transformer managed by the TNY279 is not enough to start the CM6807.
+
+
+The application circuits are as follow:
+
+.. image:: {filename}images/ZR24W/tny279_app_circuit.svg
+   :alt: Example application circuit for the TNY279PN controller (from the `datasheet <TNY279PN>`_)
+
+.. image:: {filename}images/ZR24W/CM6807_app_circuit.svg
+   :alt: Example application circuit for the CM6807 controller (from the `datasheet <CM6807>`_)
+
+In this PSU, the input DC of the TNY279 (the point just before R5 in the app
+circuit) in connected just behind the PFC circuit, i.e. just after the D2 diode
+of the CM6807 app circuit.
+
+In order for the PSU to generate the 5V rails even when the CM6807 is off,
+there is derivated DC input path, from the bridge rectifier to the input DC
+rail, consisting in a diode (D922 on the PCB) followed by a thermistor (R915):
+
+.. image:: {filename}images/ZR24W/input_DC.jpeg
+   :alt: Input DC showing both paths (yellow: via the PFC, red: the alternate
+         path when power saving is on).
+
+
+Finally, the culprit
+--------------------
+
+While testing the PSU powering the CM6807 from an external source, I notices
+another strange behaviour: the TNY279 enters a failure as soon as I sink
+current from the 5V rail, but it remains faulty as long as I let the CM6807,
+thus the PFC running, and the input DC at 400V (instead of 318V when the PFC is
+not activated)..
+
+At this point, I though there is not way the problem can come from the 5V
+regulation loop, neither from the CM6807 circuit, so the only culprit left
+would be either the TNY279PN or a component close to it, a cap (especially the
+BP/M one), a diode or a resistor. So I tested these parts (again, to be fair),
+and I replaced the cap connected to the BP/M pin. With no improvement, the only
+remaining faulty part must be the TNY279PN.
+
+So I decided to buy a couple of them from RS, and 2 days later I replaced it,
+which did fix the PSU.
+
+What a tricky half-failing part!
+
+
+
+
+.. _TNY279PN:
+   https://www.power.com/sites/default/files/product-docs/tny274-280.pdf
+.. _CM6807: http://www.championmicro.com.tw/datasheet/Analog%20Device/CM6807.pdf
+.. _FP6185: http://www.feeling-tech.com.tw/km-master/ezcatfiles/cust/img/img/24/fp6185drv02-g1.pdf