Hantek HDG2002B AWG: 5Mhz or 100MHz? Let's see!

[fremen67]

Anyway, I presume most of the changes will be on the FPGA side of the board.  Do you have the ability to do JTAG boundary scans?  It would be good to compare 1002 vs 1004.  The UCF file for the 1004 revision is here: https://github.com/alexforencich/hdg2000/blob/master/fpga/fpga.ucf .  Can you possibly check this against your board and see what matches?  Don't bother with the DDR2 part, this should be fixed due to the embedded memory controller.

I won't be able to try it till wednesday. I never did it before but with TopJTAG it should not be too difficult (well I suppose...). Could you give me the pinout of JP3?

[alex.forencich]

Anyway, I presume most of the changes will be on the FPGA side of the board.  Do you have the ability to do JTAG boundary scans?  It would be good to compare 1002 vs 1004.  The UCF file for the 1004 revision is here: https://github.com/alexforencich/hdg2000/blob/master/fpga/fpga.ucf .  Can you possibly check this against your board and see what matches?  Don't bother with the DDR2 part, this should be fixed due to the embedded memory controller.

I won't be able to try it till wednesday. I never did it before but with TopJTAG it should not be too difficult (well I suppose...). Could you give me the pinout of JP3?

2nd post on this page: https://www.eevblog.com/forum/testgear/hdg2002b-awg-firmware-reverse-engineering/

JP3: JTAG for U5
From marking end, +3V3, TCK, TDI, TDO, TMS, GND

Xilinx has a 'standard' 1x6 JTAG pinout, but this is a different pinout. 

[Howardlong]

I just received one of these units and have updated the system.inf successfully.

I have a couple of questions if I may, I haven't found the following addressed, but apologies if I've missed it.

a) In the calibration procedure, are the AC and DC measurements taken with a 50 ohm load?

b) Again in the calibration procedure, should I be entering peak to peak or RMS values?

c) Has anyone had any luck with the Word Generator? I doesn't seem to save my values for each index, and the interval entry seems a bit flakey.

Many thanks, Howard

[Howardlong]

OK I answered my own a) and b) questions by looking at thr behaviour of the device at the already calibrated low frequencies. And, it wasn't that well calibrated!

Firstly I did not use a 50 ohm termination because that wasn't the way the device was already cal'd. I also realised very quickly that it's not a very accurate 50 ohm impedance either, even at DC.

For the first ten or so DC measurements I used a DVM, so far so simple.

For the gazillions of other measurements I used an Agilent 54831d scope with a calibrated 10x probe coaxially fitted _directly_ into the AWG's output to avoid any standing waves problems due to non-termination at higher frequencies. I used the mean Vpp measurement on 10 averaged waveforms, and the measurement was reset each time, with the time base and vertical set appropriately as necessary.

It took a long time, probably an hour or so just for one channel, but the results are excellent compared to how it was.

However I've now realised there's a hardware filter tweak I possibly ought to have done first.

By the way, make sure the AWG was set to sine wave on the channel in question before calibrating. Trying to calibrate on a square wave is no fun due to the overshoot when not terminated, and the cal procedure did not seem to reset this on my unit.

[IanJ]

Question about calibration..........

Am I right in saying you just hook up a volt meter to it's output and enter in all the voltages that appear at the output. It seems to throw a few voltages to me to start with but then did a whole load the same, even negative voltages till it completed.
When I then tested......my amplitude is all over the place at various frequencies.

Am I missing something with the procedure?

Ian.

[alex.forencich]

Question about calibration..........

Am I right in saying you just hook up a volt meter to it's output and enter in all the voltages that appear at the output. It seems to throw a few voltages to me to start with but then did a whole load the same, even negative voltages till it completed.
When I then tested......my amplitude is all over the place at various frequencies.

Am I missing something with the procedure?

Ian.

Use a scope.  The unit needs calibration points at various frequencies to compensate for front end frequency dependence. 

[fremen67]

OK I answered my own a) and b) questions by looking at thr behaviour of the device at the already calibrated low frequencies. And, it wasn't that well calibrated!

Firstly I did not use a 50 ohm termination because that wasn't the way the device was already cal'd. I also realised very quickly that it's not a very accurate 50 ohm impedance either, even at DC.

For the first ten or so DC measurements I used a DVM, so far so simple.

For the gazillions of other measurements I used an Agilent 54831d scope with a calibrated 10x probe coaxially fitted _directly_ into the AWG's output to avoid any standing waves problems due to non-termination at higher frequencies. I used the mean Vpp measurement on 10 averaged waveforms, and the measurement was reset each time, with the time base and vertical set appropriately as necessary.

It took a long time, probably an hour or so just for one channel, but the results are excellent compared to how it was.

However I've now realised there's a hardware filter tweak I possibly ought to have done first.

By the way, make sure the AWG was set to sine wave on the channel in question before calibrating. Trying to calibrate on a square wave is no fun due to the overshoot when not terminated, and the cal procedure did not seem to reset this on my unit.

You did not let me enough time to reply 
I can't add a lot more as I did the same way:
- sinus wave
- no 50 ohm termination
- a DVM for the 9 first values
- My MSO5062D for the 3 sets of 27 values from 500kHz to 95Mhz, using average value on 16 samples
- directly fitted in the HDG output
- also took an hour for the first channel
- gave up for the moment for channel 2
The results are now very accurate for DC values after calibration.
For AC it is better as I can now go up to 100Mhz but I can see the impact on a change of output range on the HDG. There are small steps on the output after each change of range. I just run a self calibration on the MSO before calibration but my probe must not be very accurate.

Regarding Wave generator, it was not working at all with FW 1.00.1 so I gave up and just used the software on the PC for that. I did not test it since I upgraded to FW 1.00.2

[fremen67]

Question about calibration..........

Am I right in saying you just hook up a volt meter to it's output and enter in all the voltages that appear at the output. It seems to throw a few voltages to me to start with but then did a whole load the same, even negative voltages till it completed.
When I then tested......my amplitude is all over the place at various frequencies.

Am I missing something with the procedure?

Ian.

From what I remember, only the 9 first values are DC. You need a scope for the others ( 3 sets of 27 values I think)

I recommand that you write down all the values you measured at the same time you enter them... just in case you have to start again from the beginning

[IanJ]

Question about calibration..........

Am I right in saying you just hook up a volt meter to it's output and enter in all the voltages that appear at the output. It seems to throw a few voltages to me to start with but then did a whole load the same, even negative voltages till it completed.
When I then tested......my amplitude is all over the place at various frequencies.

Am I missing something with the procedure?

Ian.

From what I remember, only the 9 first values are DC. You need a scope for the others ( 3 sets of 27 values I think)

I recommand that you write down all the values you measured at the same time you enter them... just in case you have to start again from the beginning

Yep, managed to suss it out........great stuff!......and now I have a reasonably even amplitude for all freq up to 100MHz.....rather than tailing off severely not much above 20MHz.

I'll go over it again more accurately now I know how it's done.

Ian.

[andrija]

I can confirm telnet over LAN works fine once you add the necessary parts and configure the (static) IP address. So you don't have to get a FTDI if you don't have already and can "upgrade" to 100MHz by editing the file via LAN. Just enter "root" as username (there's no password).

[alex.forencich]

I can confirm telnet over LAN works fine once you add the necessary parts and configure the (static) IP address. So you don't have to get a FTDI if you don't have already and can "upgrade" to 100MHz by editing the file via LAN. Just enter "root" as username (there's no password).

Really, you can telnet in as root without a password?  That's a bit dangerous. 

[andrija]

I don't think it's dangerous, this is just an instrument, not a PC or a router or a NAS. Besides password would have to be known anyway in order for the service to be of any use. Technically, they should disable the service out of the box and force you to open it and use UART to start telnet or sshd. But let's not forget they don't actually give you LAN anyway, you have to buy and solder a bunch of stuff onto board so it's hardly a real issue. So running service is really a non-issue, I don't think even legally speaking they would be liable.

[Cyber7]

Telnet wasn't enabled on my 1.00.2 unit; the daemon command was remarked in boot.sh, so a pure LAN upgrade will not work on newer units by default.

Yes, there is no password for root. If you pass on the shell as a parameter you can even avoid the login prompt completely, just like the serial TTY interface. 

[andrija]

Ah so they did the right thing in the next version. I got mine a week ago so there is really no way to know what you'll get. boot.sh is where they specify start up daemons? There is no /etc/init.d/* structure? I guess it really is bare bones linux.

[Cyber7]

There's just a single init.d/rcS script that in turns calls boot.sh
It's stripped down for embedded use; for example, only ubifs and fat is baked into the kernel.

Code: [Select]

#!/bin/sh
#This is the first script called by init process

mount -a
mkdir /dev/pts
mount -t devpts devpts /dev/pts
mdev -s
echo /sbin/mdev > /proc/sys/kernel/hotplug
source /etc/profile
mount -t ubifs ubi1:config /config
/dso/etc/boot.sh &

[Howardlong]

Folks

Can recommend a source of DM9000AEP LAN chips? The usual Digikey/Farnell/Mouser seems to draw a blank, they are listed but no stock.

Cheers, Howard

[alex.forencich]

Folks

Can recommend a source of DM9000AEP LAN chips? The usual Digikey/Farnell/Mouser seems to draw a blank, they are listed but no stock.

Cheers, Howard

ebay

[Howardlong]

Ok, thank you, ordered.

[rosasharn]

Hello!
A newbie for LAN. 
Anybody has a simple manual about how to soldering DM9000AEP LAN chips and how to adjust software?
Thanks!

[Cyber7]

For device info with images see: http://hdg2002b.simplesite.com/412252681

For the soldering basics, watch Dave's video, which includes QFP packages like the DM9000:
Essentially, after fluxing all pins& pads, and tacking 2 opposing pins in place, you can a) flood all pins with solder, then use a broad iron tip over a copper wick, and remove the excess, or b) use a fine chisel pointed tip, apply some solder to the tip, and work 2-3 pins at a time, cleaning the tip to allow surface tension to aid in clean up when pins bridge. Use the tip like a paint brush-like motion pulling way from the pin. I personally used method (a), as it is faster. Be careful with the existing caps near the device; you don't want to shift or bridge them.

If you have access to a temp controlled hot-air gun with a pencil tip, see this video: Note the paint brush-like motion pulling way from the pin when touching up with a regular iron!

If you think you've screwed up and can't get solder to wick up from behind pins, don't panic! Add some more solder & Use a broader tip with the wick; a 3mm tip may seem like overkill for method (a), but it works fine. Do not increase the temp. If nothing works, buy some ChipQuick lo-temp rework solder, apply liberally around the entire device. Use a temp-controlled hot air gun to remove the device. You can use a soldering iron if you use a blade tip & move quickly. After removal, clean all pins and pads with the solder wick method. Inspect for bridges. Start over: clean with alcohol, re-flux, solder again.

Tips: Practice!  Buy some cheap AVR/PIC QFP's and some test pcbs to practice. Clean all surfaces with 99% alchohol and kimwipes first, then use non-cleanup flux (chipquick). Do not use needle point tips, as you need consistent heat at the tip; use conical or tapered tips. Temp should be <= 650F, but be sure to keep the tip moving; do not linger. Apply solder to the wick when starting to aid in heat transfer.

Inspect your work with a magnifying glass in bright light, or preferably a bi-ocular microscope with 15-70x mag.

There's no software adjustment required. A kernel module is already present in the bootup sequence. Telnet is mapped to busybox. You could cross-compile other network daemons if you really wanted ssh or other functionality. Keep in mind that aside from a debug interface, there is no other network support for the device at the moment.

[markone]

Hello,

just put my hands on HDG2002B purchased on ebay (PCB 1004, SW 1.00.2, FPGA 20, KB 3)  and i have important problems like :

- FM modulation does not work, playing with parameters often let the device to BSOD
- sometimes BSODs  are followed with a screen asking confirmation for a recovery to default image (that clears the config file hacking)
- no way to maintain screen intensity parameter to the chosen value, it changes continuously
- no way to obtain a correct "last status" recovered at boot time
- gorgeous jitter plus short glitch (about 20ns) before the rising edge (depending on the signal) on sync signal output
- rust on internal shielding edges

I'm wondering if i got a lemon or if this instrument is the pile of **** that it seems to be is not ready to be sold at current FW development stage.

[flatlander]

Quote from: Stevie1966 on September 29, 2014, 01:34:30 PM

I have resolved this problem 3 days ago.
You can try the following.
Take away the capacitor C231 and C208. Then change L25 and L17 to 0 ohm.
Wish you luck. 

On my board (PCB: 1004) these changes, to remove excessive ringing from the outputs, were already done ex factory. They also removed C209 and C232.

[Richard_L]

Further to markone's post - is anyone following this thread actually managing to use the Hantek without it freezing or BSODding on a regular basis? From reading through all the posts, it does seem prone to tripping over its own shoelaces when basic waveform parameters are changed, even without the software hack being applied or the hardware being modified.

Also, and it's probably nothing, but from looking at the specs of the HDG20XXB range, there are some parameters that don't improve incrementally with each model - something which might be expected with a pure software implementation of specification limitation, they're bunched into pairs instead, which might perhaps indicate aggregating a few small hardware changes as the models get more expensive?

http://www.hantek.com/en/ProductDetail_149.html

on the parameters tab... see the pairing on:
Frequency characteristic --> arb.
and
Pulse --> pulse

If there's anythingto this, it would suggest that the 20MHz model is the first one that's got tweaked hardware.

I'm torn between buying the HDG2002B and one of the Siglent SDG10XX series. Despite the higher price and lower spec of the Siglents, I'm erring towards one of those because of their apparent greater reliability. The Shenzen & Hong Kong based ebay sellers of the Hanteks say they offer a 3 year warranty but I'm not sure I'd like to test that in a years time.

[markone]

The Shenzen & Hong Kong based ebay sellers of the Hanteks say they offer a 3 year warranty

mmmh, i wouldn't bet on it ...

[andrija]

Further to markone's post - is anyone following this thread actually managing to use the Hantek without it freezing or BSODding on a regular basis? From reading through all the posts, it does seem prone to tripping over its own shoelaces when basic waveform parameters are changed, even without the software hack being applied or the hardware being modified.

I referred to this as a toy when speaking to my friends and I still do. I certainly had my own share of BSODs but the device is useable. I can't trust its amplitude much after calibrating it and the non-sine waveform is not all that clean after a few MHz so I'd say, think of it as a much better alternative to those cheap DDS generators from EBay. I would certainly go for Siglent if you want something more reliable. I've been trying to buy a used HP/Agilent as an outdated but reliable and precise reference function gen but for some reason prices on EBay have soared in the last two months to laughable levels - you may as well buy a Rigol 4k series for the price they want for those old HPs. I wonder why - did someone do a popular video on one of these recently? Anyhow, if you want a toy this Hantek is fun, if you want something that you can rely on and not fuss over, get something else.