Re: HA8SA All Band CW QRPp DC Transceiver
Thanks for sharing the details Pete. The ESP32 platform is by far my homebrewing favorite these days for price, power and flexibility; from the Xiou -S3, to the D1-Mini and the DevKit like this one.?
73, George N2APB
|
Re: HA8SA All Band CW QRPp DC Transceiver
That's an interesting design. The TX/RX switching is clever.? Did you build it, Pete?
73 Mike M.? KU4QO
"A very simple direct conversion HF CW transceiver. It works on any band just need to replace LPF circuit or use a 30 MHz LPF and external tuner. On 40m I've had 17 DXCC on Europe. I had only 225 mW and reach London and Moscow too. During ARRL 10m contest I was very lucky with propagation because of I made contact two USA stations with 71 mW. I checked the out power more times because I did not want to believe. I was not wrong it was really 71 mW on 10m."
?
?
YouTube:
?
?
Schematic
?
?
?
Esp32-DevKitC?
?
?
?
?
|
Re: Which frequency governs the Si5351 initial phase offset calculation?
Hi Elia!
where is the phase offset added to the signal? Before the R divider or after?
Conceptually, think of the 2nd multisynth dividers as counters that can be initialised with a certain start value that?represents the phase offset; they all start counting with the initialized?value when the PLL is reset, and hence flip their CLK outputs with corresponding?phase offsets. It is the vco that drives the 2nd multisynth diver, which is followed by a static R divider.
In practice under the hood of the si5351 there is much more ongoing; SiLabs (now Skyworks) describes the multisynth as [1]:
"The first stage of the MultiSynth architecture is a
fractional-N divider, which switches seamlessly between the two closest integer divider values to
produce the exact output clock frequency with 0 ppm error. To eliminate phase error generated
by this process, the MultiSynth calculates the relative phase difference between the clock
produced by the fractional-N divider and the desired output clock and dynamically adjusts the
phase to match the ideal clock waveform. This novel approach makes it possible to generate any
output clock frequency without sacrificing jitter performance."
So, this means that besides fractional frequency division there is something in-place that actively (phase) corrects the momentary jitter that? otherwise fractional division would create. This was also mentioned in a recent paper that analyzed the si5351 fractional?divider [2], and here the assumption is that there must be some kind of 256-step phase interpolator in-place near the vco that performs these corrections. Looking into an older SiLabs patent [3], figure 11 seems indeed to confirm this, but this would also mean that the actual vco frequency is actually oscillating at 5 GHz, and not in the range 600..900 MHz [4] as the datasheet would let you believe. It is probably also this phase interpolator that is used to program the phase offsets.
73, Guido
[1]? [2]?
[3]?
[4]?
On Saturday, May 10, 2025, Elia DL1YE via <elia= [email protected]> wrote: Mike, sorry for the late reply, I'm rather busy at work at the moment.
?
Thank you for confirming that it is the combined divider ratio from VCO frequency to output frequency that determines the required phase offset.
?
The fact that the division ratio reduced the phase offset is not intuitively clear to me, but I obviously have the wrong mental model of how the synthesizer architecture works.
I'm curious if anyone knows how the internal architecture achieves the delay?
?
If the delay were to come right after the VCO output before the input into the multisynth and R divider combination of CLK1, the 90 degree phase relationship between CLK0 and CLK1 should be kept regardless of divider ratio as long as it is identical for CLK0 and CLK1.
Why would CLK_PHOFF be specified in terms of 1/4 of the VCO period if the delay doesn't directly follow the VCO?
?
Thank you!
?
73, Elia DL1YE
|
RF Power Transistors (2SD882, 2SC5706) more tests
I did more tests with both transistors today, adding 30M and 20M. Same setup as before, switching to the output transformer for the 4 and 5 watt tests, but then I thought why not check at the 3 watt level with and without the transformer--in most cases the gain was a little higher with the transformer. I didn't think of doing that until the 30M test, so the 40M data doesn't have that. Neither transistor will output 5 watts but I'm again surprised the lowly 2SD822 even puts out a little more at 20M than the 2SC5706.
Probably not many folks have much interest in any of this but I had fun doing the tests and it's easy to post the data.?? :)
73,
Steve AA7U
|
Re: Which frequency governs the Si5351 initial phase offset calculation?
Mike, sorry for the late reply, I'm rather busy at work at the moment.
?
Thank you for confirming that it is the combined divider ratio from VCO frequency to output frequency that determines the required phase offset.
?
The fact that the division ratio reduced the phase offset is not intuitively clear to me, but I obviously have the wrong mental model of how the synthesizer architecture works.
I'm curious if anyone knows how the internal architecture achieves the delay?
?
If the delay were to come right after the VCO output before the input into the multisynth and R divider combination of CLK1, the 90 degree phase relationship between CLK0 and CLK1 should be kept regardless of divider ratio as long as it is identical for CLK0 and CLK1.
Why would CLK_PHOFF be specified in terms of 1/4 of the VCO period if the delay doesn't directly follow the VCO?
?
Thank you!
?
73, Elia DL1YE
|
Re: RF Power Transistors (2SD882, 2SC5706)
Thanks Steve. That's particularly good news about the 2SD882's from Tayda - especially for transistors that only cost 14 cents apiece. I recently received delivery of 10 of them. Will need to add a few more to my next order, for the parts bin and potential future projects.
?
Dave
AA7EE
|
RF Power Transistors (2SD882, 2SC5706)
Here are some test results I did with the 2SD882 (from Tayda), and the 2SC5706 (from kitsandparts.com) on 40M. I used 12.8 V supply, and the Norcal 40A buffer, driver and power amp circuit with a 40M LPF with two harmonic trap capacitors. The Rutledge book "Electronics of Radio" uses these circuits for several Problems for the student--he drives the circuit with a function generator so that's how I did mine--that works very well plus you can record the input level for different output wattages, and compare it with other power amp transistors. I used a J309 for the buffer and a 2N3904 for the driver; I added an extra 2N3906 keying transistor, same as used on my earlier RF Amp Test Fixture and not the Norcal keying circuit. The Norcal 40A output drives 50 ohms directly and is suited for 0-3 watts output. I also tested for 4 and 5 watts output, changing the 18 uH RF choke on the output collector to the usual 1:4 step up transformer. I've attached the schematic for all this.
The bottom line is both transistors work very well as single-transistor power amplifiers at 7 MHz, both giving up to and including 5 watts output from the 12.8 VDC supply. No heat sinks were used; very briefly keying the circuit allows for this. Both transistors, being quite small with small heat sink area, immediately begin to heat at key-down and power immediately begins to drop, so I have to catch the initial peak voltage across the 50 ohm dummy load on the Siglent oscope. The Rutledge book shows how to measure power amp transistor gain, converting the base RF voltage to dBm and subtracting that from the output wattage/dBm. I checked at 2 watts, 3 watts, then changed to the output transformer and checked at 4 watts and 5 watts--doing this for both transistors. I was surprised that the 2SD882 would give 5 watts output, at least on 40M; and gain was still holding up at 5 watts for both transistors. Note the phasing dots on the driver transformer--this is important (of course also shown in the official Norcal 40A schematic, and in the Rutledge book--I just followed the schematic). I experimented with reversing the secondary phase, monitoring the output power set to 2 watts with the proper phasing, and the output decreased by 0.1 watts when the phasing was reversed. NA5N's "Data Book for Homebrewers and QRPers" also mentions this, see section around Fig. 4-7.
Chuck Adams has already mentioned to the list about the 2SC5706 being used in Diz's 5 watt transceiver and separate 5 watt amplifier, both use balanced pushpull output for the 5706's and cover up to 30 MHz. Of course proper heat sinks are used.
73,
Steve AA7U
|
Thanks Allison for explanation. I had heard that small signal characteristics were same, so that explains it.?
Before the internet I recall an engineer asking a colleague for a PNP complement to a particular NPN he wanted to use. I realize hole and electron behavior is different, so I wonder what parameters they were seeking to match?
?
in early days there were countless transistor part numbers for small signal types. I wonder if this way due to process variation. Or to mimic the countless tube part numbers. Often most any silicon NPN will work in many circuits.?
Curt
?
?
?
|
HA8SA All Band CW QRPp DC Transceiver
"A very simple direct conversion HF CW transceiver. It works on any band just need to replace LPF circuit or use a 30 MHz LPF and external tuner. On 40m I've had 17 DXCC on Europe. I had only 225 mW and reach London and Moscow too. During ARRL 10m contest I was very lucky with propagation because of I made contact two USA stations with 71 mW. I checked the out power more times because I did not want to believe. I was not wrong it was really 71 mW on 10m."
?
?
YouTube:
?
?
Schematic
?
?
?
Esp32-DevKitC?
?
?
?
?
|
note the 2n2219A is the TO5 can version but the same die as 2n2222A.
The difference is more power better cooling with heatsink.
?
I use a lot of the 2n2222A (aka the roach) as its one of the oldest silicon
transistors still manufactured.? IT was created in 1961.
?
Allison
|
Re: RF Power Transistors (2SC2078 used by Rick Choy in Midway Electronics kits)
Hello,
Looking at my ME40+ kit instructions.? It says it can be either a 2SC2078 or 2SC2166.
?
Mine is a C2078.? I already have a transistor tester, but I have seen some home brew tools posted on here lately that I want to try.?? But like you said, this transistor should work fine for the ME40+ kit that I got since Rick has already tested them.
?
The main reason I wrote this message is - has anyone considered other transistors like the one available at kitsandparts like the 2SC5706?? Or that transistor found on Tayda D882??? As long as that transistor would not require a change to the circuit or how the ME40+ radio works? Would it??
?
I know Rick has worked hard to give us this ME+ kit faithful to its original design.? Thank you goes to Rick!
?
I do look forward with the updated ME+ kits return as Mark I and Mark II.? Especially the other bands and the Mark II.??
?
Thoughts on other transistors?
?
73 Daniel KK4MRN
?
|
Re: RF Power Transistors (2SC2078 used by Rick Choy in Midway Electronics kits)
Hi Rick,
OK, I'll call your bluff.?? :)???? I have your ME20? I built but
it has a 2SC2166 in it which I won't pull and test (which you
didn't mention using), but your ME40? I built has the 2SC2078 in
it--I pulled it just now and put it in my RF transistor power amp
test fixture which has a 20 meter LPF (with two harmonic trap
caps),? and runs at 12.8? VDC; I use a function generator to drive
the test fixture (I've shown the circuit in previous postings). I
test for a solid 5 watts output,? 45 Vpp across a measured 50.06
ohm 12 watt dummy load. I've found that a good RF transistor will
give 5 watts output at 12.8 VDC supply. I put your 2SC2078 in the
test fixture and varied the function generator output to get the
maximum output that I could get, at 14.0 MHz. The most I could get
out of your 2SC2078 was a paltry 25.6 Vpp, 1.64 watts! So your
2SC2078 is just another fake counterfeit just like all the others
I've tested from eBay and Amazon so far (and from Dan's Small
Parts and Kits) except for the one seller which I've mentioned, in
Las Vegas. A genuine 2SC2078? (from old CB pulls) that I have
several of will output 46-47 Vpp at 14.0 MHz; the good Chinese
2SC2078 from the Las Vegas seller will output 45.5 Vpp or a little
more.
I think you need to revise your testing procedure if you claim
the 2SC2078 that's in my ME40 is not defective and is not a fake
counterfeit. But since your kits are only designed to put out
1.5-2 watts, I suppose these fake counterfeit transistors you're
using will do the job if you drive them hard enough.
73,
Steve AA7U
On 5/8/2025 4:09 AM, R. Choy via
groups.io wrote:
toggle quoted message
Show quoted text
I use the 2078 in my QRP kits I sell at Midway Electronics.
The ME40, ME80, ME30, and ME20. I get all mine from ebay
sellers. They are name branded but that doesn't mean anything. I
individually test each final transistor for power out, spectral
purity and heat dissipation. So far of the 400 or so I have
purchased, I haven't come across one that was defective. YMMV
Rick
|
Re: RF Power Transistors (2SC2078)
I use the 2078 in my QRP kits I sell at Midway Electronics. The ME40, ME80, ME30, and ME20. I get all mine from ebay sellers. They are name branded but that doesn't mean anything. I individually test each final transistor for power out, spectral purity and heat dissipation. So far of the 400 or so I have purchased, I haven't come across one that was defective. YMMV
Rick
|
Re: RF Power Transistors (2SC2078)
As of a few minutes before 10pm Pacific Time, he has now sold 55 and has 45 left. I wonder how many of those sales were as a result of Steve's posts in this group? I wouldn't be surprised if it was nearly all - or perhaps all of them!
?
Dave
AA7EE
|
Re: Benchtop DMM Recommendation
Thanks for the recommendations. It looks like either of those would be a great instrument to have on the bench. I have been watching some reviews and will likely make my choice in a day or two.
73 , Jerry W0PWE
|
Re: RF Power Transistors (2SC2078)
The seller of the good 2SC2078's I tested shows 80 of them now available--his new shipment of them must have arrived. He's raised the price slightly to $2.10/each.
toggle quoted message
Show quoted text
On 4/28/2025 5:11 PM, Steve Ratzlaff wrote: My last purchase of 2078s was from another US seller on eBay--his ad shows testing them on a Peak meter--he tests them for beta (gain) at the low currents any of these test meters can only test at. I corresponded with the seller, saying such a test at such low currents didn't tell much, that a real test would be for output power. He responded that he could only test the way he was showing but that previous purchasers seemed to be happy with these 2078s. I bought 5 of them to test--they finally arrived today, after 12 days! He's in Las Vegas, only a day or two mail time from me in Arizona,? so for unknown reasons he appears to be very slow in shipping. I immediately tested all 5 and I'm very happy to report they all passed--3 gave 46 Vpp output, the other two 45.7 and 45.6 Vpp--exceeding the reference 5739 set for 45 Vpp. So these are the first Chinese 2SC2078s I can recommend, if you're looking for a decent RF power transistor. is the item and seller. Again note that you'll probably wait quite awhile to get these.
73,
Steve AA7U
|
60 meter homebrew rig: TX amplification demo
I made a video short on youtube for a part of of my? proposed 60 meter transceiver. The video shows the TX power amplifier and small signal amplifier boards working together under test. Please note, I'm no video producer ;) This is a variant of the amplifier boards I used in my 40 meter transceiver project with a the necessary low pass and bandpass filter modifications for 60 meters.
?
Here is the video short of the amplifier under test:
?
https://www.youtube.com/shorts/blvfe2LJ2Lw
?
Using the 2 boards together, I can apply -45dBm of drive with the signal generator and get just over 40dBm (10W) of output. Between the two boards, there is 85dB of total amplification.
?
The two stages on the TX preamplifier board are MMBT3904 and 2SC3357. These are classic feedback amplifiers as outlined in EMRFD. This board also as a triple tuned bandpass filter on its output to clean off any unwanted mixer products. This board provides 40dB of amplification.
?
The? two stages on the power amplifier board are push-pull.: 2 2SC5706's followed by 2 Mitsubishi RD06HHF1's.? The TX low pass output filter is a 5th order chebyshev. I'm seeing -67 dB of attenuation on the second harmonic, and the third is below the noise floor.? There is also a TX relay and a receive output for TX-RX switching. This board provides 45 dB of amplification.
?
|
Curt,
I have been a ham since 1960 and recently met and friended an 86 year old gentleman who became a ham just 3 years ago. ?We are having a ball. He just discovered the Morserino and convinced his wife of 60 years to participate in learning CW together. ?What a blast!!! ?A life long learners permit IT IS!
toggle quoted message
Show quoted text
On May 6, 2025, at 10:13?AM, Curt via groups.io <wb8yyy@...> wrote:
? Dave
?
A ham license is indeed a life-long learners permit. Glad for all your learning and extensive sharing with us. Our community is so strong because of so many folk sharing their experience. Of course after seeing a project by AA7EE or AA8V many of us don't want to visually share our work (-:
?
I was wondering not long ago about effectiveness of heatsinks on those classic TO39 1 to 2 watt transistors. Practical answer is if that heatsink gets hot, it's working. I will evaluate that metal slab on my tut80 when I have it open. I recall the tut80 has better audio for my oscillating ears than the nc2030, so I will check out those schematics.?
?
Nice work by Steve to share his work. RF lower NPN are scarce but not extinct. Maybe the one installed in my ugly weekender is original, if so I haven't destroyed any yet. I have killed off numerous JFETs.?
?
Curt
?
|
Dave
?
A ham license is indeed a life-long learners permit. Glad for all your learning and extensive sharing with us. Our community is so strong because of so many folk sharing their experience. Of course after seeing a project by AA7EE or AA8V many of us don't want to visually share our work (-:
?
I was wondering not long ago about effectiveness of heatsinks on those classic TO39 1 to 2 watt transistors. Practical answer is if that heatsink gets hot, it's working. I will evaluate that metal slab on my tut80 when I have it open. I recall the tut80 has better audio for my oscillating ears than the nc2030, so I will check out those schematics.?
?
Nice work by Steve to share his work. RF lower NPN are scarce but not extinct. Maybe the one installed in my ugly weekender is original, if so I haven't destroyed any yet. I have killed off numerous JFETs.?
?
Curt
?
|
Re: Benchtop DMM Recommendation
If you are ok with 50MHz scope bandwidth (more like 30MHz in reality), you can go even lower in price:
?
It works fine as a DMM. I have an HP34401A to compare it with. There is a few counts of error for currents. Voltage is even better, 1-2 counts of error. Resistance is about +-0.5% (I have some 0.1% resistors and again comapring to HP34401A).
?
73, Mike AF7KR
|
George N2APB
hwstar
David Wilcox K8WPE