LowPowerLab Forum

Hardware support => RF - Range - Antennas - RFM69 library => Topic started by: joelucid on September 03, 2015, 04:13:42 AM

Title: RFM69 & HopeRF: are they Semtech clones or not?
Post by: joelucid on September 03, 2015, 04:13:42 AM
Ouch!   :o  That's a rather serious bug!   :'(   Does that bug exist in  just the RFM69x knock-offs, or is it a bug in the orginal Semtech too and all of the Semtech clones?   :-\

I don't think the Moteino's use knock-offs. The problem might be related to my use of a non interrupted burst sequence rather than a sequence of individual packets. Still it's definitely a bug. You can workaround by having the radio accept all packets and letting the 328p decide which ones to take. That should work as long as there isn't a naturally occurring signal that's close to the burst sequence.
Title: Re: RFM69 & HopeRF: clones or not?
Post by: emjay on September 03, 2015, 04:44:01 AM

For the benefit of the community, perhaps you could identify these "RFM69x clones" you refer to. You have personally tested some? How you distinguish "clone" from original?
Title: Re: RFM69 & HopeRF: clones or not?
Post by: emjay on September 03, 2015, 10:18:22 AM

In summary, is it fair to say you have suspicions that the RFM69 series modules do not use Semtech silicon?
Seemingly based on:

The implication is that the integrators of RFM69 modules are somehow complicit in this IP abuse.
I'd suggest you clearly identify your remarks as speculation until you have something more substantive to report.

Component procurement from the Far East is indeed a potential minefield - the outcome of your continued research will be illuminating.

Title: Re: RFM69 & HopeRF: clones or not?
Post by: Felix on September 03, 2015, 11:43:05 AM
... my loose understanding is that the Semtech design was the "original," and that the RFM69x, and possibly others by other manufacturers, are copies.  Otherwise, how do you explain why their datasheets are practically identical?  A different explanation I've heard is that maybe they all licensed the same cores from yet a different party altogether. Or perhaps HopeRF is just a module maker, and perhaps it is using genuine Semtech chips.  I wish I knew enough to say for sure which of these scenarios might be the case.    I'm a Noob when it comes to the RFM69x, so I could be completely wrong.  If so, please do correct me. So far I've only managed to catch glimpses of what may or may not be going on, such as the "special note about hopeRF" written at the bottom of the page at this link: https://openrf.codeplex.com/  In particular, what it says is, "HopeRF is a Chinese module manufacturer that offers low cost modules based on a number of various manufacturer's products.  Their datasheets are typically fully of errors and their technical support is non-existent. The modules require external filtering to be compliant with FCC regulations in the US.  Furthermore, we have found the modules to have quality issues.  For these reasons,  Digital Six Laboratories and the OpenRF™ project are no longer affiliated with HopeRF in any way.

Perhaps this doesn't belong here, but since you asked, and to set the record straight about RFM69 and HopeRF.
I hope this clears up the confusion fog of assumptions and baseless rumors on the internet:

The HopeRF RFM69 modules use genuine Semtech chips. I have decapped these chips to verify this. They are custom packaged hence you see a Hope logo and "RF69" markings on them and not the default semtech markings. Take the following facts:

- if HopeRF would be using a cloned silicon design, and since Semtech owns patents and the IP they and would sue and stop imports in the civilized world where laws actually still work (US, EU). So Hope would have no market there
- the packages for HopeRF's RF69 chips are QFN28, perhaps a little cheaper to package in China or Thailand/Taiwan. I bet it's a magnitude cheaper to buy the silicon or straight out wafers and doing  the custom packaging with subcontracting, allowing Hope to make the chip appear proprietary. That's great for HopeRF and for all the rest of us, we get the same genuine chips, at a low price, win-win for all.
- the Semtech packaging is QFN24
- the HopeRF RFM69 radio variants are based on the stock Semtech reference designs found on pages 76/77 of the datasheet (http://www.semtech.com/images/datasheet/sx1231h.pdf)
- the reason (implied) for HopeRF's copycat datasheets is that they are not selling the chip itself, but a reference design based on the original chip. So of course they have to change the datasheet to reflect these changes and pinout diagrams, and in doing so they slapped their logo on it (if Semtech doesn't complain why would anyone else?). It's unfortunate that there are some errors in the datasheets but if one of us had to do the same and translate a chinese datasheet with your own changes added, I wonder how well we'd do. What bothered me somewhat is that they just replaced "Semtech"
- the concept of fake not only scares but annoys me. If these were fake and buggy and not performant modules I would not use/resell or endorse them, instead I would make my own. In fact I made a batch of RFM69 just for fun to prove myself I can do it. I bought Semtech chips and followed their reference design which is very close to what HopeRF did. But to get close to make it worth it for me to keep making these I would have to make them in 10,000qty and spend half my time doing it. Hence I prefer to source from HopeRF, that's all they do on a massive scale and they do a fine job. It's one of the Chinese products that is of good quality. Thumbs up from me, I hope they don't screw up in the future.
- internet statements claiming HopeRF modules are low quality pivot on the fact that some of the SMD components are reflowed at dodgy angles making it look like the design is sloppy. I see something else: - they use the same PCB for half a dozen variants of RFM69 and hence most pads are smaller than optimal for reflowing and some pads are shared, so some pads are not always identical or perfectly square with others etc. So yeah - at that scale reflowing will pull the parts wherever there is more surface tension. Perhaps an optimization rather than a compromise and another reason the price is so "low".

Notice the Semtech SX1231H chip markings and the HopeRF RF markings:
(https://farm1.staticflickr.com/584/21088513486_b5e2ea546f_c.jpg) (https://flic.kr/p/y8w9m7)

Regarding this statement:

... For these reasons,  Digital Six Laboratories and the OpenRF™ project are no longer affiliated with HopeRF in any way.
You will find they make their own Semtech based products, USA made, high quality FCC certed, but closed source and AT command based - which is great for commercial integrators, so a slightly different market than what Hope targets. So I think the bias in this statement is completely justified ;), but a bit unfair on the part of HopeRF, given the fact that modular FCC certification is one thing and overall product FCC certification is another, which in the end would be required for any end consumer product whether it uses Hope's modules or Digital6's or anyone else's.
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: Greg_Wasik on January 08, 2016, 11:54:12 AM
Semtech lists Hoperf as one of their partners.


Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: OSBSS on February 03, 2017, 04:48:38 PM
Thanks for the detailed clarification, Felix.

I do have a question though: You said that making these modules yourself and selling them would be worth it if you buy 10,000+ Semtech chips and would take a lot of time to do so. I was wondering - why not incorporate that SX1231H circuit into the Moteino board itself instead of buying modules from HopeRF? Pardon my ignorance, just wondering if it would be advantageous to do so. Perhaps smaller overall footprint, about the size of the Pro Mini, or perhaps some cost savings? I would assume pick and place on just one side of the board would also save a lot of time as opposed to both sides and reflowing twice (if that's how it's done)
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: Felix on February 04, 2017, 08:10:07 PM
The modules are manually soldered on the bottom side.
The modular approach allows for a lot of flexibility and lowering the volumes and variants of boards assembled at one time.
I think the only scenario where an integrated RF solution makes sense is where it can be made in very high volumes.
Also there's the technical aspect of designing the board as a whole to suit the RF circuit (for compliance and impedance matching), which is not a very simple task.
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: perky on February 04, 2017, 08:21:07 PM
It's going to be pretty tough to beat the HopeRf module pricing at high volumes though, you'd need some pretty heavy purchasing power to overcome the component and assembly cost of a discrete solution IMO. You might just overcome it at very high volumes I suppose, but they'd have to be pretty high.
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: OSBSS on February 06, 2017, 02:37:27 PM
Ah, yeah, that makes sense, re: compliance and impedance matching.

But these modules are manually hand soldered on each Moteino? They are very neatly done! Does sound like a lot of work though.

Just thinking of automation - if these are pick and placed, can they be reflowed again with the rest of the board?
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: Felix on February 06, 2017, 02:45:56 PM
Thanks, I've been doing it for years and many many times, I think it's down to a science now :), only takes a few seconds.
There are 2 aspects in terms of pick & place:

- reliably picking them - the pick head must grab hold from something (ideally in the center), none of the modules have a good center grab surface. My P&P uses mechanical centering (optical also but i don't like to use that since it's much slower to move to an up looking camera in a corner then back to the board) while quickly moving the part towards place position, so the RFM69 is not a good part for picking or would require lots of tweaking to avoid it flying off the head
- RFM69 being custom packaged by HopeRF in a QFN28 which has no specification. This packaging they use could very well be a very cheap one (which I'm sure is part of the reason they just buy wafers and package it themselves) which could absorb moisture and then upon reflow the moisture could damage the die inside (without necessarily an immediate and evident effect). So to ensure that would not happen it would require baking the parts in a special oven at a lower temperature to evaporate any moisture, and only then allow it for placement/reflow. I doubt anyone else does this, and Im not eager to, I just prefer hand soldering, it allows for much more compact boards overall
Title: Re: RFM69 & HopeRF: are they Semtech clones or not?
Post by: OSBSS on February 06, 2017, 03:00:02 PM
I need more practice :) Honestly thought it was either hot air or PnP. Cool stuff.

Okay this makes sense. Thanks for explaining. So back to OP - these are 'genuine' chips, but some corners are cut to get down to an affordable price at low or high volumes.