GL823 is not working well

[remopong]

Hi, I'm trying to use a GL823 since a few months. I almost make it work but, it doesn't... Let me explain:

I wired it up as I think it should be. My windows 10 detects card reader, but not the microSD i'm tring to reach.

Please help me. I'll provide my schematic in an hour or a more.

Any help is welcome...

[remopong]

Here's the schematic:

[mon2]

Your schematic is not complete or the GL823K is not wired up correctly.

Can you clarify the power rails for this IC as per the pin details from the datasheet:

https://www.electrodragon.com/w/images/c/cb/GL823K.pdf

VDD value ?

VDDA value ?

5V pin connected ?

[remopong]

Hi, my bad, the screenshot I posted is wrong on some wiring, let me send it again...

I was the datasheet but do I have to connect the VDDA and VDD or are them just outputs ?

The 5V is connected to the USB as in the new schematic. I bought a micro SD reader from inrternet and opened it. As I knew, it uses a GL823 too. I guess I'm just gonna test every pins to see how they are connected to the SD

Thanks, I'll let you know if I have anything new...

[remopong]

I manage to link those, thanks to a multimeter. I'll check if it works

[Ian.M]

There's a reference design from Genesys "GL823 SCH_QFN24B_200.pdf" attached here: https://www.eevblog.com/forum/projects/gl823-sd-card-reader-no-response/

Although its not the same chip (being the 24 pin version) its worth noting that it uses SD 4 bit mode, with four data pins, CMD and CLK all connected to corresponding pins of the card.   It also shows VDD linked to VDDA via a ferrite bead.   

What reason do you have for believing that the GL823K will stay in 1 bit/SPI mode once its detected and initialized the card?  Why do you think the GL823K can use its CMD pin as MISO?  Why do you think it can put the card in SPI mode with CS tied low, when one would normally apply 74 or more clock pulses to initialize the card before taking CS low?

Have you checked both VDD and VDDA have 3.3V on them? If not, it will need a ferrite bead linking them like the GL823.

[ttt]

I don't see any decoupling caps or bulk cap for the 3.3V internal regulator. You will likely need that.

A bit of googling reveals this schematic: https://www.pianshen.com/article/64011166102/

[remopong]

I can see some capacitors, am I forced to use them as well ?

If yes, what are they doing is this case and do I have to use the same capacity else it wont work ?

Oh and my chip is a ssop16...

My PMOS is currently at 2.09V instead of 3.3V


Thanks.

[remopong]

I added an ams1117 with its capacitors (pcb built-in capacitors) to provide a good 3.3V to the GL823. So: I have a stable 3.31V on the VDD and PMOS.
 
My stepDown looks like this one: https://mbatechmeds.com/product/ams1117-3-3v-power-supply-module-mountable-on-pcb/

At the beginnig my pc detected the reader dark grey, now its light grey. Something happened but I still cant access it...

I don't really understand the LED though, sometimes it blinks fast, sometimes it just stays ON...

[Ian.M]

So what made you decide to back-feed VDD?  If you looked at the datasheet, you'd *KNOW* the GL823K has an internal regulator and doesn't have a pin to disable it, so should never be back-fed.   The design Ttt linked doesn't have a regulator, and nor does the GL823 reference design in the topic I linked.

If you suspect or know that your SD card is going to draw more current that the GL823K 3.3V regulator would be happy providing via the PMOS pin without the voltage dropping, the correct approach would be to use an external 3.3V regulator just for the SD card, with an active high enable pin driven by GL823K PMOS (+ a pulldown).

Try building the design Ttt linked exactly as drawn (with the exception of possibly omitting the ESD protection U7 SMF05C), with C158-161 as close as possible^* to the GL823K and C9 as close as possible to the SD card socket.   Also see "GL823 Design Guide_100.pdf" in the topic I linked, as its nearly all applicable to the GL823K.

Decoupling and its placement is *IMPORTANT*!

[remopong]

I managed to make it work... I'm really fu**ing happy right now.

I tried differents capacitors on VCC and GND of the sd card, because I dont have a lot of capacitors at home, i put them in parallel, right now I have 2*470 uf and when I plug everything in, after 15 seconds or so, the sd is finally recognized...

Any lower capacity only makes the sd card to reset due to the (i think) lack of current

 Thanks everyone !

[Ian.M]

470uF (let alone two in parallel) is grossly excessive and risks damaging the GL823K's internal PMOS switch by excessive inrush current.  They may also cause problems for the host USB port.  Try 100nF disc, resin encapsulated or SMD ceramic (or anything ceramic from 10nF to about 220nF).

[remopong]

Why would they damaged the usb port or the chip ? I have now 1000uF and it works perfectly too...

Why would applying less would work better ?

I don't have any ceramic, only chemical through hole capacitors now...

The PMOS is also connected to air, should I link it to anything ? I can't link it to VDD though

[Ian.M]

Ah, I was assuming you were still supplying the SD card with switched 3.3V via the PMOS pin. If you have it fed from the AMS1117, that regulator's internal overload protection would limit the inrush current into the capacitors, so there is nearly no risk of upstream port damage (till the AMS1117 dies from repeated abuse, and possibly shorts Vbus).

Ceramic capacitors have much lower ESR and self inductance than wound construction electrolytics, so are more effective for decoupling, especially at higher frequencies.

IMHO, you should order a selection of ceramic caps to hold in stock.  Even free next day parts delivery from a good distributor, is no substitute for the convenience of only two minutes to select a cap from the box on the shelf.

[remopong]

Thanks,

I'm planning on using the AMS1117 but I can see a lot of people talking about tantalum capacitors, am I forced to use them ? Or I can use chemical / ceramic ones ?

Somewhere it's written 10uf on each side (input and output) elsewhere it's written something else... Datasheet says: dividing bridge + diode + 2 tantalum capacitors....

My circuit is gonna be ON for several hours per days almost everyday... Do I need very good components or something ?

[Ian.M]

You can often get away with the parallel combination of an electrolytic and a ceramic capacitor that's a small fraction of the total value in place of a tantalum capacitor.  If you need something more like a tantalum, look at solid polymer aluminum electrolytic capacitors, which have two major advantages - they don't contain manganese dioxide electrolyte (a strong oxidant) so don't act as miniature incendiary devices as they fail (though they can still overheat and even catch fire), and aren't made from a conflict mineral.


N.B. The various revisions of the USB standard limit capacitance directly on Vbus in any device to max. 10uF to limit the inrush current and thus erosion of connector contacts, so don't go significantly above 10uF on the input side of your regulator.

[remopong]

Thanks again, I'm now working on the PCB itself.

What I am placing in input ? Ceramic or electrolytic though ?

[Ian.M]

Probably the best option would be 10uF 16V ceramic with 0.1 ohm resistor in series (with the capacitor, not the supply) to add some ESR to provide damping + 100nF 16V ceramic across the lot to improve its decoupling effectiveness at higher frequencies.   
However you may be able to get away with a 25V 10uF ceramic on its own (the higher voltage rating to reduce the loss of capacitance with voltage, which if you use too low a voltage rating capacitor can easily combine with the lead inductance to cause resonant peaking exceeding the regulator's abs. max. input voltage rating even if that's several times the nominal supply voltage, possibly destroying the regulator).

Normally, I'm cheap and don't really care about extra lifespan as long as its good for several years, so a 10uF 'wet' aluminum electrolytic in parallel with a 100nF ceramic will do the job.

[remopong]

Well yeah several years should do the trick, so this (image attached) is not good enough ?

https://lcsc.com/product-detail/Multilayer-Ceramic-Capacitors-MLCC-SMD-SMT_Samsung-Electro-Mechanics-CL31B106KOHNNNE_C47971.html
(10uf 16V for both input and output)

[Ian.M]

It depends - I don't see any figures in that capacitor's datasheet for capacitance loss with respect to DC bias voltage for that size and voltage X7R capacitor.  From other sources, I know its likely to be between 1/3 and 2/3 of its nominal capacitance at its max. rated voltage, and even if it didn't loose any value, resonance with the lead inductance could cause the input voltage to peak at 10.5V (double the max. permitted USB Vbus of 5.25V) during the ringing after the device is hot-plugged.   If it looses 1/3 of its capacitance, the ringing could easily peak at over 16V, enough to kill an off-brand 1117 regulator. 

Also, %DEITY% only knows what ringing voltage the GL823K can withstand on its 5V pin, as that datasheet is a joke, with no abs. max. ratings whatsoever.

[remopong]

So I should go for a at least 25V for sure ?

[Ian.M]

... or an electrolytic in parallel with a ceramic of 10% (or less) of its value, as the ESR damps the ringing.

https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf goes into preventing regulator failure due to ringing over-voltage in much more depth. 

[remopong]

Ok, so. I added a 1k ohms resistor to the input / gnd of the ams1117 so that the capacitors cant discharge when I unplug it. When I don't, if I try to replug my device, it's not recognized by windows.

I can't find any help or information on what does the blinking led means... When I plug my device, the LED blinks for 15 seconds before my computer can recognize and access the SD card... This is really annoying to wait to this long...

[arekm]

You seem to have experience with GL823, so I have a question.

Can GL823 be prevented from accessing sd-card? There are some GPIOs but can't find any information about possible usage of these nor any information about USB endpoints provided by gl823.

Background:

There is a device https://github.com/FYSETC/FYSETC-SD-WIFI (schematic available there) which goal is to provide wifi access to sd card plugged into some other end device.
It also provides usb-storage device to host (like pc computer) over microusb. When this device is connected to usb host it automatically acts as card reader (usb-storage provider).

I'm looking for a way to tell it explicitly to provide usb-storage to pc host OR to do not do that and just leave alone sd card allowing end device to freely use sd card.

There are more expensive devices (https://wiki.tizen.org/SDWire) with that capability but I wonder if it would be possible to modify code or hardware of existing fysetc card to behave in similar way.

GL823 behaviour is the key.

[remopong]

Hi, pardon me for my extremely late answer, I'd say that reading back what we all wrote, the best way would be to power off the GL823 to avoid it accessing the sd card.. At least that's what I was doing at the time in this project..

Hope you'll read this !