Posts by G8UGD

    Hello Toni;

    I will try to answer the best I can in my own words on why I think it is better, I can not prove it or disprove it. The modification information from SP8XXN and SP5XMU mention the board is a Doherty amplifier. In reading up on these amps (wiki style) one device is switched on in class AB and the other is basically class C and off, needing a high peak level to switch it on, as it is used for the high peak to average transmissions of UMTS.


    The article I read also said that power is lost as the input power is split two ways into each device by a 90 degree phase shift with different lengths of track, so in my mind half the input power is being diverted to a device that will not amplify any signal until it reaches it's designed peak switch on point. I have no idea at what level this is!


    So to me, to get back this wasted input drive the second device needs to be biased as the first, in a class AB mode.


    Now as it was designed for around 2100 MHz the 90 degree phase shift is not going to be 90 at 2400Mhz as the tracks are longer then required, but any phase shift change on the input is being reversed on the output or close to it.


    As to drive level, I know that my pre-driver starts to limits out at less than 2 Watts, I have also reports that others that have done the mods have achieved a very maximum of 120 plus watts for short periods and this is not something I would NOT recommend, I have taken mine to an excess of 90 watts before my nerve gave out.


    There are others doing similar mods, in fact a James Smith posted on facebook and in in IO groups

    https://groups.io/g/DATVPA/top…_amplifier_using/34475663?


    James is probably the first that brought it to my attention followed by a local radio ham Tim G4WIM

    PA für 2,4 GHz


    Each has done it a slightly different way but end up with similar results. I have been to Tims's location and know his test gear is accurate. Do try and have a look at both links.


    Do not expect to run the devices at full power and for them to last long also do not forget you need something like a peak to average difference of 6dB, but if you know that the peak output is say 80 watts that is 3dB better than saying the peak output is 40 watts and the average is 20 watts rather than 10 watts.


    I hope this explains it from my perspective, it is probably the best I am capable of giving. :)


    Adrian

    OK thanks

    I was looking for something to help a couple of new starters get correct alignment of the dish/dual feed etc. Mintiuner needs time to lock and provide an MER as does my GT-Media sat finder, so something that could see all 8 plus MHz of the band and not rely on high processing power when pushing an SDR over the range. Something will crop up to mind at some point. Thanks for the heads up on future development.


    Adrian

    I should have been more descriptive, looking for a scan style app that uses the tuner of the Minitiouner and displays signals it receives, similar to a spectrum analyzer.

    The QO100_Live_Tune is taking the BATC WB server and using that as an aid to set freq and SR rate on the Minitiouner, is relies on a internet connection. That is a good app for fast tuning and works well, but not what I am after at the moment.

    I have modified the two amp's I have, I have removed the input circulator and the input 2.2Ghz filter but left the output circulator in place, I am glad I did as I burnt out the cheap Ebay 50 watt dummy load at some point and did not realise. So the circulator would have been taking all the load in effect.


    Going off my circuit above I have done a few mods in changing the 1K1 resistor to a 1K0, removing the 250 Ohm to ground and adding another 4K7 to ground all on the right hand side of the board. I have also cut the track feeding the emitter follower and fed in to another point. The end effect is that both devices now have the same bias voltage.



    You may just notice where I cut the track under the end of the white wire!

    Is it the best way of doing it, probably not, it is probably better individually biasing each device for a set standing current that way you know each device is taking the same current. Out put powers have been in excess of 80 watts. The other mods have been snow-flaking, cutting tracks and removing an input cap as per some posted mods. Do be careful, the more you push these the more prone they can be to self destruct.

    OK, went a different route to 'import sys' as I found a tutorial on another method which is probably a lot safer.


    I am up to this at present:-



    #!/usr/bin/env python2
    # -*- coding: utf-8 -*-
    ##################################################
    # GNU Radio Python Flow Graph
    # Title: dvbs2
    # Generated: Sat Aug 3 00:17:48 2019
    # Modified for external command line variables G8UGD 26/11/2019
    ##################################################

    from gnuradio import blocks
    from gnuradio import dtv
    from gnuradio import eng_notation
    from gnuradio import filter
    from gnuradio import gr
    from gnuradio.eng_option import eng_option
    from gnuradio.filter import firdes
    from optparse import OptionParser
    import osmosdr
    import time
    import argparse

    parser = argparse.ArgumentParser(description='Frequency, IF gain and Symbol rate')
    parser.add_argument("-cf", default=2409.7, type=float, help="This is the 'cf' centre frequency variable")
    parser.add_argument("-ifg", default=1, type=int, help="This is the 'ifg' TX IF Gain variable")
    parser.add_argument("-ks", default=250, type=int, help="This is the 'ks' Symbol rate variable ie 250, 333 etc.")
    args = parser.parse_args()
    cf = args.cf
    ifg = args.ifg
    ks = args.ks


    class dvbs2(gr.top_block):

    def __init__(self):
    gr.top_block.__init__(self, "dvbs2")

    ##################################################
    # Variables
    ##################################################
    self.symbol_rate = symbol_rate = ks * 1e3
    self.taps = taps = 80
    self.samp_rate = samp_rate = symbol_rate * 2
    self.rolloff = rolloff = 0.35
    self.resample = resample = int(round(5e6/symbol_rate))
    self.center_freq = center_freq = cf * 1e6
    ##################################################
    # Blocks
    etc


    I have never programmed in Python so I am sure there are better ways at doing this, Changing the modulation and FEC seem to be more difficult, but I will continue to try. It keeps the mushy grey matter healthier.


    Adrian

    Sorry to ask, but I could do with a little bit of help to get me moving.


    I use gnu radio to drive an SDR for DATV, I slightly reworked an existing flowgraph without any GUI, and use the generated .py file from a script in Linux.


    I am wondering if I can pass variables from command line to the python file at run time.


    For example here are the first few lines from the Python script.


    #!/usr/bin/env python2
    # -*- coding: utf-8 -*-
    ##################################################
    # GNU Radio Python Flow Graph
    # Title: Sr250
    # Generated: Sat Aug 3 00:17:48 2019
    ##################################################
                                                                                                                                       
                                                                                                                                       
    from gnuradio import blocks
    from gnuradio import dtv
    from gnuradio import eng_notation
    from gnuradio import filter
    from gnuradio import gr
    from gnuradio.eng_option import eng_option
    from gnuradio.filter import firdes
    from optparse import OptionParser
    import osmosdr
    import time


    class SR250(gr.top_block):

    def __init__(self):
    gr.top_block.__init__(self, "Sr250")

    ##################################################
    # Variables
    ##################################################
    self.symbol_rate = symbol_rate = 250e3
    self.taps = taps = 80
    self.samp_rate = samp_rate = symbol_rate * 2
    self.rolloff = rolloff = 0.35
    self.resample = resample = int(round(5e6/symbol_rate))
    self.center_freq = center_freq = 2408.9e6

    ##################################################
    # Blocks
    ##################################################
    self.rational_resampler_xxx_0 = filter.rational_resampler_ccc(
    interpolation=resample,


    So it would be good to add the frequency as an external variable and possibly the symbol rate. I think the answer lies with Import sys, but to date my tries have failed me.


    So if there is a Python Guru out there that could put me in the correct direction it would be appreciated.



    Adrian

    Hello Matthias, Thanks for the information. I am using my MHL21336 to drive the Bisonelectronics amp. It seems to work OK for now until I feel the need to get something better. Very similar results on output to yours. I do not have a VNA the gear to test mine fully, but it seems to be OK.


    Adrian

    Apart from the disadvantage of having more than one Co-ax going to it, holes through the wall etc. You could run one co-ax from vertical polarisation to a receiver for SSB and one co-ax for horizontal polarisation for DATV stuff, so in theory you can have two receive systems at the same time, there is no need for a 18 Volt PSU or buck up converter to get the voltage from a standard 12-13.8 Volt psu.


    Typically if you have a 4 co-ax/connector lnb, it gives you the opportunity to modify it and feed with an external gps locked source for ssb RX and still have a V and H output.


    All that being said I still run a standard un-moddified single port lnb. :)


    Adrian

    Thank you for your response, it is appreciated. One thing that still puzzles me is the length of the 'dipoles'


    I get one leg of the dipole to be 35.3155mm for the too long and 28.013mm for the too short. these seem to be a lot longer and a lot shorter then 1/4 wave at 2.4GHz, I accept that there could be a change to the electrical length of the dipoles as they are close to 'ground' as it where. But even if I consider no shortening, I get the short length would be resonant at around 2.675GHz and 2.12Ghz for the long length. When I looked at the instructions for the POTY it showed theoretical resonances less than 35MHz above and below 2.4GHz.


    Put me down as being a pain in the A**, I like to understand what is happening rather then just take it for granted, so I can ask a lot of questions.


    Adrian

    OK thanks for that, not sure if any clearer to me. I accept what you say, that is not an issue, I would guess that the different lengths from the feedpoint to each 'element' would also add a phase change as one would have a shorter path from the feed therefore the other lags behind, but that suggests to me they are not 90 degrees out. Not to worry I think if I get chance i will search for some simple theory on patch antennas. Guess I am only used to 1.4 and 1/2 wave style dipoles.

    :)

    So I have been straining over a magnifying glass and doing searches on the internet for SMD markings. If anyone with the same amp would like to verify my findings, I would appreciate it.


    I have drawn out a bias circuit, sorry it is not in Ki-cad or similar as I do not yet know how to use circuit programs.

    V104, V107 and V201 are marked 1PN which I believe to be a NPN transistor similar to FMMT2222.

    V108 is marked as W2F 2d. Which I believe is similar to PMBT2907A and a PNP transistor.

    V105 is marked R1E and appears to be a 1.2 Volt fixed shunt regulator as in a LM4041.

    Lastly V106 marked RAC again a LM4041 type device this time adjustable shunt regulator.


    The Green circles are my idea of ground points, resistor values in purple and what voltages I measured in Blue.

    The Analog Device AD5259 I have reduced to a variable resistor in the diagram between 1.2 Volts and GND.



    Not sure what is going on with V107 as it seems to be just using a diode junction.


    Anyway the point for me is understanding the biasing of the power devices, with the left hand one being at a measure 1.9 Volts and the right hand side one at 0.57 Volts.


    Adrian

    Just throwing a question out there, I guess the answer will be no, but does anyone have a circuit diagram for these amps I would love to follow the bias circuit. My eyes are definitely not good enough to follow the components and tracks on the board to figure it out. There seems to be some checking that the peak amp (classC) is powered before it allows bias to the class AB? device and have noticed a track that runs around the bottom of the board for this, wondering about using this track for bias of the normally class C device but would need to work out the switching part first.


    Or does anyone know where/what the boards came out of, i.e. equipment model number etc, and I would try and give some company a message request.


    Adrian

    I would be very doubtful of getting a good VSWR at 2.4GHz by using resistors in parallel, at these frequencies dummy load resistors tend to be co-axial in nature. Have a look at this page http://millimeterwave.free.fr/ATT_DISPONIBLES.htm and see if anything takes your fancy, it really depends on if you can deal with 7/16 connectors. Also you will need more that 40 dB in total to not damage your power sensor. A quick look on Ebay (UK) and I see some 50 Ohm loads 25 watts for about £11, they will only be rated for intermittent use I would suggest, but they say they are good to 3GHz or even some resistors that can be bolted directly to a heatsink that will take a lot more power if cooled correctly, so please do have a look around various outlets before you decide.

    I am probably guessing wrongly, but wonder if you will be trying to test a Pyrojoe 20 watt amp or similar, if you can give your self some margin in case it is 25 or 30 or even more Watts output.


    What ever you do be careful at these frequencies and powers as you are taking microwave frequencies now and soft tissue damage is always a possibility.

    Very carefully! with the aid of something like a Ebay 30 dB directional coupler, a 30 dB attenuator and a 50 Watt dummy load good for 2.4GHz. The directional couplers seem to be very good are are covered by a few posts on the forum, they cover 800 to 2500MHz and are cheap to buy, I picked up a 50 Watt dummy load for around £30 UKP and you can get a N connector'ed attenuator for similar money, so probably not a cheap option but one which could save your power meter.


    Amp through directional coupler to the load, -30dB port to another 30 dB attenuator to your power meter, so if you assume 43dBm with your 20 Watts you will get a signal at the power sensor of +43 - 60 = -17dBm

    Adrian