Review: Hantek DDS 3X25. Anyone own one?

[Mechatrommer]

more test...
yellow=synch out
blue=hantek out

b_o = synch out (notice the glitch)
b_q = square out (multiple at 1mhz, 1.37mhz = out of multiple)

[Mechatrommer]

b_r = ramp output at 24&25mhz, 25mhz got in multiple.
b_s = sine output at 30&40mhz, u should already know which is not in multiple.
pls note mhz is megahertz, i lazy to type "shift" key.

[Mechatrommer]

b_t = trapezia at 10,12,20 mhz.
pop quiz... which one is in multiple?

[Mechatrommer]

conclusion... if you need both synch and out port, really you need to check whether you are in multiple or not before doing your circuit, otherwise you'll lose hair. if you dont understand, ask. if not, dont buy Hantek 3x25
ps: the good news is... this model is obsolete.

[saturation]

Are you terminating both the sync and signal output of the 3x25 with 50 ohms, as its needed?

[Mechatrommer]

err, no only the synch out got terminated. whats the effect? i only can think of somesort of loading stuff

[saturation]

Yes, if not properly terminated on both outputs, the values are unstable at that high frequency.  Second the sync signal is only good to 25 MHz, which its what its rated for, above it is also unstable. 

You can generate sine wave with reduced amplitude to 75 MHz, but you can only sync internally.

As you can see my outputs are no where looking like yours.

Yellow = Output
Blue = Sync out
Purple = FFT at rectangular, clean to 25 MHz, no visible harmonics at this scale.

Its phase shifted, as I wrote in this thread, but the signal is cleaner.  It also shows the importance of the right cables and terminating impedances.  I don't have the ringing you have in the other posts regarding making cables.

Enjoy!  The skills in analog electronics will last a lifetime, its worth putting effort into, it hasn't changed in ~ 100 years.

err, no only the synch out got terminated. whats the effect? i only can think of somesort of loading stuff

[Mechatrommer]

well ok, i tried my best to make similar setting as yours. made a quick fix 50ohm for the second terminator (i cannot sacrifice another bnc connector, i only have 3 left, 2 died during soldering, gotta use the dc-bnc type) still i got ringing in synch out, maybe the 75ohm impedance connector is the culprit. and i dont understand why you say you have a clean FFT for your square signal, as i understand square consists of alot of harmonics.

redo the test with both (crude) terminated, my previous conclusion still hold, still the same graphs except the output amplitude now is halved due to termination. and i swap yellow=signal, blue=synch to comply with saturation's setting. i provided only one snapshot (lazy to repost the whole series again). c_t_12mhz.jpg = trapezia (terminated) at "not multiple" 12mhz.

mind to show your cable?
ps: the time now is 9AM, and i think i'm retiring for today, will continue at 3-4PM... today. hava nice day

[saturation]

It looks better.  Your FFT on the sync output has no other spurs except expected for a 'clean' square wave, = 3rd and 5th harmonic, 75 MHz and 125 MHz.



You needn't sacrifice anything if you bought the right stuff, sorry to say.

Your device requires 50 ohms, but you have a mix of items that will cause non-ideal responses as its not terminated correctly, it was OK to get a better waveform rather than have no termination, but as you try to improve it to ideal, your chances of success will not be good.  As I see it you have a 75 ohm cable, 75 ohm BNC connectors, 50 ohm resistor of unknown composition, I hope its not carbon composite or wire wound, its not the ideal terminator.  The proper resistor is a SMT type thin or thick film.  And if you have long leads sticking out like you have in the photo, it will cause unpredictable responses.




http://books.google.com/books?id=zpTnMsiUkmwC&pg=PA2&lpg=PA2&dq=carbon+composition+resistors+parasitic+capacitance&source=bl&ots=B4EZfXzDc0&sig=geOY4nFYqO7RziVEOsQrWVwT5Ko&hl=en&ei=Nnh_TfyBMMmD0QGem72ECQ&sa=X&oi=book_result&ct=result&resnum=1&ved=0CBsQ6AEwAA#v=onepage&q=carbon%20composition%20resistors%20parasitic%20capacitance&f=false

My clean FFT is for against the sine wave, and it refers to your image here, not sure why you have so many spurs although this is FFT at the dB view:


I'd be more than happy to show you my cables, but its just generic from Jameco.com and MPJA.com.  50 and 75 ohm cables and connectors cost the same amount of money, so buying 75 ohms instead of 50 ohms well, is a path down the wrong road.

The bright side, is its all a learning experience for a better future.  When I get hold of my scope, I'll duplicate what you did in the photo with the 12 MHz triangle wave.

well ok, i tried my best to make similar setting as yours. made a quick fix 50ohm for the second terminator (i cannot sacrifice another bnc connector, i only have 3 left, 2 died during soldering, gotta use the dc-bnc type) still i got ringing in synch out, maybe the 75ohm impedance connector is the culprit. and i dont understand why you say you have a clean FFT for your square signal, as i understand square consists of alot of harmonics.

redo the test with both (crude) terminated, my previous conclusion still hold, still the same graphs except the output amplitude now is halved due to termination. and i swap yellow=signal, blue=synch to comply with saturation's setting. i provided only one snapshot (lazy to repost the whole series again). c_t_12mhz.jpg = trapezia (terminated) at "not multiple" 12mhz.

mind to show your cable?
ps: the time now is 9AM, and i think i'm retiring for today, will continue at 3-4PM... today. hava nice day

[saturation]

A 50 ohm terminator, similar to most folks use, about $2 each.  You connect these to one end of a split connector, such as the T.


T connector, about $2.50 each.  Insure they are 50 ohm types.


These are $10 each, and its straight through.  Cost more than the setup above.  There maybe some advantages to the shape of this connector at very high frequencies, but the above works for most purposes, particularly within the range fo the Rigol 1052e and the 3x25.

[Mechatrommer]

well, whats the purpose of the last one?

[saturation]

Its one piece and since its straight, it may do better in highest frequencies.  For high frequencies you need to avoid sharp 90 deg bends in any conductors [like the T connector], it can induce capacitances that can alter Zo, but practically speaking I think you can use T or Y adapters to over 300 MHz.

See under 'routing'.  
http://www.icd.com.au/articles/emc.html

For more info a google search term is signal integrity. For one reference that has everything, see 'High Speed Digital Design.'

well, whats the purpose of the last one?

[saturation]

I presume this is trapezoid waveform, yellow, and blue is sync signal.  Not sure which frequency used, so my screen captures are for 12.5 z.

The FFT is to show that the harmonics present in the waveform is to confirm its trapezoidal, not sine as they look very similar.  My mistake, your previous images with the harmonics explains that.  I'll correct my reply.

[alm]

50 ohm resistor of unknown composition, I hope its not carbon composite or wire wound, its not the ideal terminator.

What's wrong with carbon composition for this purpose? Before alternatives like thin film came around, this was actually the resistor of choice for high-frequency work. Wire wound can be OK if it's wound with an Aryton-Perry winding (cancels most of the inductance), otherwise it's obviously horrible. Even the extra ~10nH of a carbon/metal film resistor shouldn't be an issue, and I wouldn't expect major issues with 50/75 ohm mismatch either. I agree with your other points, though, and in my opinion this setup looks more suitable for connecting a light bulb than a high(ish) frequency signal. This is not DC or audio (which is basically the same), you can't just splice two wires together and expect good signal integrity.

[saturation]

Mecha has found a new bug in the 3x25. I think I understand now what he was saying by 'multiple' and 'leap year trick'.   I had time to duplicate some of his tests and he discovered a new problem with the sync signal.  Its exactly what he describes but I couldn't understand what his metaphor was.

Its part of dajones original complaint that isn't solved completely by fixing the impedance.

At certain frequencies the sync and the Voutput remain unsynchronized!  There are however, frequencies that you can synchronize but this bug can certainly throw people off.  This is was mecha was referring as 'multiple'.  In between certain values the sync signal and the output signal are not identical, so you cannot sync both waveforms until you reach the 'multiple'.

This doesn't affect Vout, but it does affect using the sync signal to lock in Vout, if you choose.

That's why he couldn't sync at 12MHz but I can at 12.5 MHz.

[saturation]

Hi alm,

I've never really experimented with using carbon composite for anything above 1 MHz. I should give that a whirl and report back if I still composites around.  Heard of a type of induction canceling resistor that must have been it, but I've never seen one:  
http://en.wikipedia.org/wiki/Ayrton-Perry_winding

When composite was popular, I'd suspect 'high frequency ' was at a lower range compared to today.  

In mecha's tests, given harmonics of the waveform used, and the sensitivity of the sync to output impedance, I'd strongly suspect cabling plays a bigger role in the waveforms appearance but the 3x25 design flaws are responsible for the lack of synchronization. Viewing each channel using a different trigger and the right impedance produces a relatively clean waveform, see another post.

http://books.google.com/books?id=zpTnMsiUkmwC&pg=PA2&lpg=PA2&dq=carbon+composition+resistor+frequencies&source=bl&ots=B4EZgRvC7Y&sig=qJiDRN3VxUrlkxH2Rb2aQ5kBtLU&hl=en&ei=PASATfuTEsSM0QGujIX7Dw&sa=X&oi=book_result&ct=result&resnum=10&ved=0CFgQ6AEwCQ#v=onepage&q=carbon%20composition%20resistor%20frequencies&f=false

50 ohm resistor of unknown composition, I hope its not carbon composite or wire wound, its not the ideal terminator.

What's wrong with carbon composition for this purpose? Before alternatives like thin film came around, this was actually the resistor of choice for high-frequency work. Wire wound can be OK if it's wound with an Aryton-Perry winding (cancels most of the inductance), otherwise it's obviously horrible. Even the extra ~10nH of a carbon/metal film resistor shouldn't be an issue, and I wouldn't expect major issues with 50/75 ohm mismatch either. I agree with your other points, though, and in my opinion this setup looks more suitable for connecting a light bulb than a high(ish) frequency signal. This is not DC or audio (which is basically the same), you can't just splice two wires together and expect good signal integrity.

[Mechatrommer]

yes it was dajones try to highlight it in not so comprehensive report/post. i'm yet to figure out whats the basic 'multiple' of this to get "synh out" and "signal" synchronized. and for my non ideal cabling, i think at the same time it will prove how sensitive and stable the 3x25 to any slight impedance mismatch or imperfection in the circuit design. i still highly suspect its the software issue, but then to re-open and check how the hardware perform is my next "to do" list. but the easiest so far i think is to find this "basic multiple" so to work the 3x25 reliably with "synch out" port as if it turn out that this is a software issue, i still incapable to access/mod the spartan (big) chip.

recap...

The Hantek DDS 3x25 is a solid function generator with arbitrary waveform capability, far better in performance than most analog function generators, in its price range or with similar specifications.  It one of the lowest cost 25 MHz capable DDS function generator on the market, and likely the only one with AWG capability.

Man, I don't understand how you can say that!
The thing glitches like crazy for square waves above 2.5mhz ( except a few stable spots like 10mhz ).
The sync out line has no timing relationship to the output signal except at those rare "stable" spots.
The sync out line glitches every 20us ( 50khz ) and will glitch any output waveform other than sine waves.
I could NEVER recommend this to anyone, except maybe as a source of sine waves, and as long as you
didn't need to use the sync out line at all.
If you did your stability test at one of the "stable" spots ( like 10mhz ) you might want to try it again at 10.1mhz.

Xilinx DS529 Spartan-3A FPGA Family Data Sheet
made a quick view, synch out is from IO_L01P_0 (pin 110). but what the hell, FPGA knowledge is not within my reach

[Mechatrommer]

i run a program to get the LCM... not found! ah sux! you have to chart the 'in synch' frequency manually, there is no 'multiple' anymore. and i tried to look at the 'leap year' effect using pulse trigger, they seem to occur at regular or constant interval, though i need to crank up the timescale and look carefully at those many small oscillation to see 2 consecutive 'leap year' or 'out of synch' as dajones highlighted as in his picture below. so i still highly suspect this is software issue or vhdl or verilog or whatever it is



though, 'in synch' seem to be reliable up to 2MHz regardless of any frequency you are putting in, eg 1.3, 1.8MHz etc, but jitter seem to worsen at some odd numbers.

[Mechatrommer]

here i present a steady display of:
1) 'leap year' effect @ 1.981 MHz
2) 'dajones out of synch' glitch @ 2.3 MHz
(note: unterminated synch out again )

[saturation]

A comparison of effects of better cables on the waveform you have demo'd.
Some of instability has relation to the quality of the connections, but its not as big a contributor as the Hantek software bug.  Using persistence mode.  Most striking is a negative glitch is half the magnitude.

Mecha's 10 MHz.



Saturations 10 MHz.

[saturation]

Here's a sample of the Hantek output, yellow and the sync output, blue, with the Hantek programmed for 12MHz.

They are general 'clean' but off, so its not possible to sync both signals together onto the 1052e with the standard triggers, which uses one channel to trigger both channels, or even the ext sync input.

However, if you set the Rigol to 'alternate' mode, it will trigger each channel separately.  You can now estimate the difference in the frequency between each channel.

[alm]

When composite was popular, I'd suspect 'high frequency ' was at a lower range compared to today.

Yep, it was probably only up to 500MHz or so those days, although I don't see any inherent reason why they wouldn't work above that (until lead inductance becomes an issue). I wasn't suggesting using them in new designs (they're pretty much obsolete), but they're definitely better for high frequencies than most other common through-hole types (does anyone even make SMD carbon composition resistors?), like metal film and carbon film.

http://books.google.com/books?id=zpTnMsiUkmwC&pg=PA2&lpg=PA2&dq=carbon+composition+resistor+frequencies&source=bl&ots=B4EZgRvC7Y&sig=qJiDRN3VxUrlkxH2Rb2aQ5kBtLU&hl=en&ei=PASATfuTEsSM0QGujIX7Dw&sa=X&oi=book_result&ct=result&resnum=10&ved=0CFgQ6AEwCQ#v=onepage&q=carbon%20composition%20resistor%20frequencies&f=false

You can't view this page, viewing limit has been reached. What did it say?

[saturation]

Hi alm, sorry the link didn't work.  I've copied it here.  Now we can easily get 100 MHz for home labs, and 500 MHz is not outrageously priced.  When perusing various references, the word 'excellent high frequency response' and 'poor response' for carbon composition resistors are used by almost an even split of different EE articles [ not hobbyist level links] and seem contradictory. You'll notice that on a google search.   But the authoritative references like Fink's EE Handbook and most top references do us a favor and publish relative response curves.  

Here are 2 books with viewable pages online:

http://www.amazon.com/Circuit-Design-Second-Christopher-Bowick/dp/0750685182/ref=cm_cr_pr_product_top

http://www.amazon.com/Reference-Data-Engineers-Ninth-Communications/dp/0750672919/ref=sr_1_1?ie=UTF8&qid=1300319322&sr=8-1

Maybe the contradictory opinion on carbon composition resistors is that for high frequency use, its mostly just for historical interest as its just not used for that purpose today.

When composite was popular, I'd suspect 'high frequency ' was at a lower range compared to today.

Yep, it was probably only up to 500MHz or so those days, although I don't see any inherent reason why they wouldn't work above that (until lead inductance becomes an issue). I wasn't suggesting using them in new designs (they're pretty much obsolete), but they're definitely better for high frequencies than most other common through-hole types (does anyone even make SMD carbon composition resistors?), like metal film and carbon film.

http://books.google.com/books?id=zpTnMsiUkmwC&pg=PA2&lpg=PA2&dq=carbon+composition+resistor+frequencies&source=bl&ots=B4EZgRvC7Y&sig=qJiDRN3VxUrlkxH2Rb2aQ5kBtLU&hl=en&ei=PASATfuTEsSM0QGujIX7Dw&sa=X&oi=book_result&ct=result&resnum=10&ved=0CFgQ6AEwCQ#v=onepage&q=carbon%20composition%20resistor%20frequencies&f=false

You can't view this page, viewing limit has been reached. What did it say?

[alm]

Thanks for the info, I wasn't aware of the parasitic capacitance. Most sources I've seen (eg. this and this, just random datasheets) seem to consider them superior to metal film for high frequency (thin film is obviously superior). The first datasheet recommends carbon composition for a part that has transition times of 20-30ns, not exactly GHz speed, but no 10MHz either.

I find it suspicious that the graph has only one curve for carbon composition, I can't imagine parasitic capacitance having the same influence at say 1ohm and 1Mohm. I also doubt that all carbon composition resistors are useless beyond 10MHz, lots of equipment from the seventies and earlier was full of carbon composition (wirewound was inductive, and metal film expensive), many of it can go beyond 10MHz. I'd have to check if for example the Tek TM500 series (which uses lots of carbon composition and multiple plugins go up to 250MHz) use metal film in all places that need to pass high frequency. I believe carbon comp was frequently used as termination resistor, but if they'd only work up to <10MHz, they'd be almost useless, since you need fairly long cables to even notice the transmission line effect at that frequency.

I just checked, the Tek PG-502 (250MHz pulse generator with a rise time of <1ns) uses two ~50ohm carbon composition resistors (as indicated in the electrical parts list) as switchable rear termination, with no reactive components nearby to compensate for parasitics. It would need about 350MHz bandwidth to reproduce the 1ns edge, can't imagine that the termination has a zero impedance (extrapolating from the graph you posted) at those frequencies. These were just the first resistors I checked, there are many more carbon composition resistors in there (and in similar equipment of that vintage). There are some 1% metal film resistors, but they're there because of accuracy/stability. This doesn't mean that carbon comp is superior to metal film for HF, but it does suggest that carbon comp works well beyond 10MHz.

Maybe there are different ways of producing a carbon composition resistor? That would explain the conflicting data, although I've never seen something like a 'low-capacitance carbon comp resistor'.

[grenert]

How about these MELF resistors?  Vishay claims they're good to 10 GHz+
http://www.vishay.com/docs/28718/melfhf.pdf

They say low inductance, but no mention of capacitance.