Ultimate3S PA

[Steve Haynal]

Hello List,

What do those with more RF experience think of the final PA in the qrp-labs Ultimate3S radio? Three BS170s ($0.44 a piece) in parallel for max power with 12V supply.

See #5 and #6 here. See 4.5 here.

73,

Steve

KF7O

[in3otd]

BS170 are used also in the Softrock RXTX Ensemble; I wanted to test that also but didn't find the time yet. Maybe the next weekend...

73 de Claudio, IN3OTD / DK1CG

[jimATkingstagDOTnet]

Steve

With multiple transistors, I would have thought that they would need to be matched, or biased separately.

73 Jim, G3RTD

[ZL2APV]

Getting rid of heat in BS170's is the big problem. I had a little push pull amp running with a pair of BS170's and made the heat sink by sandwiching two 40 mm square plates of 4mm aluminium extrusion and drilling 2 holes of a clearance size for the fets into one edge. The holes were drilled offset from the centre so that the flat edge of the fets were in line with the face of the plate. I lined the hole with silicon heat sink washer and clamped it together with the non offset plate turned to give a flat face against the fet flat face and the silicon rubber compressed around the curved face of the fets giving me a complete contact with the case. The amplifier fell off at 50 MHz and was about 4 dB down at 10 Metres. All the rest of the bands were pretty flat. I did not use any feedback on the amp and I am sure it would benefit from that particularly if some inductance (ferrite bead etc.) was included in the feedback path to improve the frequency response at 10 M band. Unfortunately I did not make very good measurements as it was pre HiQSDR days and I had no VNA or spectrum analyser. The output was not that good on my scope but cleaned up beautifully with a filter. I was able to drive my push pull RD16HHF1 amplifier to 25 watts with that little amplifier and think it may be possible to get 4 or 5 watts from a push pull parallel combination of BS170's but it would need some experimentation. Heat is the enemy the specs on the fets are ok.

73, Graeme zl2apv

[Sid Boyce]

Ed Murphy (GM3SBC) paralleled 4 BS170's for 5W out on the Ensemble RXTX,
the other 2 fitted below the board on the PCB pads and secured with the
screws of the 2 on top.

He reported 5W on 40m but didn't mention other bands.

I got all the bits, matched the BS170's, formed the heatsink and got no
further as other projects got in the way sidelining my Softrock/UHFSDR gear.
https://groups.yahoo.com/neo/groups/softrock40/conversations/topics/72243
73 ... Sid.

> See #5 and #6Â here
> <http://www.qrp-labs.com/ultimate3/u3mods.html#morepower>. See
> 4.5Â here
> <http://www.qrp-labs.com/images/ultimate3s/assembly2.pdf>.
>
> 73,
>
> Steve
> KF7O
>
> --
> You received this message because you are subscribed to the Google
> Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hermes-lite...@googlegroups.com
> <mailto:hermes-lite...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.


--
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[Sid Boyce]

The copper heat spreader is shaped to contact to the 2 on top and the 2
on the bottom.
73 ... Sid.

>> See #5 and #6Â here
>> <http://www.qrp-labs.com/ultimate3/u3mods.html#morepower>. See
>> 4.5Â here

[Steve Haynal]

Hi All,

Thanks for the feedback! Sorry for the delayed response but my day job has kept me very busy (a good busy) the last few days.

My interest in the Utimate3s PA is as an option for a QRPp build. It may be easy to have the space for a BS170 that can be bypassed, but if desired, it can be stuffed and a bias wired to provide 200-300 mW with an external inline LPF for the band of interest. Since most QRPp users will use a mode like JT9/JT65/CW/WSPR with fairly constant amplitude, it need not be linear. I'll keep this in mind as I layout v2.

I am very fond of the PA in the RXTX. The design looks like it would easily connect to the differential output of the AD9866. At <2W, PE4259s could still be used instead of relays. Imagine a companion PA/Filter board with all surface mount components and the RXTX PA. But this is a project for some other day... 

Below is an interesting e-mail forwarded to me from KC9OJV from W8TOD the creator of the RXTX PA.

Hi John

As you said, I have tortured the BS-170 and find them to be pretty much indestructible. They will certainly produce a reliable 1.5 watts or so with a 3 unit assembly.

I have some concern regarding directly implementing the Ultimate 3S circuit due to the bias adjustment that circuit uses. The Ultimate 3S does not require a linear amplifier and I assume the Hermes lite does. The circuit used in the RXTX has very good bias regulation and the mechanical assembly used also provides some thermal feedback that stabilizes the BS-170s in extended key-down operation. This circuit uses 2 BS-170s thermally coupled to a third which is used to regulate the bias. I have found it to be very stable, very insensitive to device selection, and very linear.

If you use a four device array play close attention to bypassing and keeping the RF connections fat and short.

The RXTX SDR is a wonderful performer, producing amazing SSB quality due to the use of stable phasing. I have also expressed concern however when using the RXTX as an exciter for higher power amplifiers. It requires careful attention to output filters in order to meet harmonic performance as well as control of the drive levels to meet IMD3 constraints. An RXTX driving an IPA and PA would produce unacceptable levels of spurs if driven “balls to the wall” and requires pretty stringent care in setting that level. The rigs also do not have ALC or any audio processing so, while the SSB produced has wonderful fidelity, it has a distinct lack of “punch” unless an external audio processor like “Voice Shaper” is used.

So, in summary, if you plan to use BS-170s:

- Be careful of the LPF and harmonic performance.

- Take care with the bias with an eye toward linearity and acceptable IMD levels with two tone tests.

- Run it from 13.8 volts or even higher for better power and linearity.

- Use external audio processing for SSB.

- Monitor, monitor, monitor…….. do not rely on signal reports for quality assurance.

It sounds like a wonderful project. Good luck and thanks for contacting me on this.

Best regards,

Warren Allgyer - W8TOD

On Sep 22, 2015, at 1:04 PM, John Greusel  wrote:

Hi Warren,

I'm on a group that working on developing a new SDR called Hermes Lite- I'm mostly a lurker as it's a pretty high level group.

It uses the same family of front end as a Hermes or the new Apache lab units so it can mimic those mature designs to PSDR and other software.

The brilliance of it is the RF side is based on a Cable Modem Transceiver.

I thought you might be interested- just google Hermes Lite and you'll find it.

Steve, the designer, mentioned using BS170's (see note below) and I said Warren has done more to study (torture) those than anyone I know.

I said I'd write to you and see if you had any comments. I believe he's talking about a preliminary amp to get the signal up to a workable level.

[in3otd]

Hello,
I have done some measurements on the Softrock Enxemble RXTX push-pull amplifier, which uses two BS170 (plus a third for bias stabilization).
In short it seems quite nice, can get about 2.5 W PEP or 1 W CW (at 10 MHz, did not measure at other frequencies), harmonics are a bit high, though.
All the measurements results are at http://www.qsl.net/in3otd/ham_radio/BS170_PA/BS170_PA.html .

[Steve Haynal]

Hi Claudio,

Thanks for doing this experiment and posting the results. I always appreciate your excellent reports. I like your use of Python and Matplotlib. I am putting together a budget spectrum analysis flow using a DS1054Z scope, PC, Python and matplotlib.

What are your thoughts on using as TH6226 as an HF driver? I am pretty excited about the possibilities and intend to try it out. Here is what I like about it:

• When used as class AB, one should be able to around 500 mW (same as Hermes with OPA2674) and use up to 15V power supply. This is a great build option for those intending to drive another amp.
• When used as class H, one should be able to get more power, possibly >2W. Efficiency is high so heat sinking requirements are reduced. This is a great build option for those intending to run a Hermes-Lite QRPp.
• The cost is reasonable in single quantities at $4.41.
• The fixed gain simplifies layout and saves space. The AD9866 should be able to drive at different levels.
• There are two complete drivers in each package, which opens up possibilities for two stages, or parallel op amps.

73,

Steve

KF7O

[Steve Haynal]

Have you shared your report with the softrock40 list. I am sure they would be interested.

73,

Steve

KF7O