Topics

ubitx - stop press

Ashhar Farhan
 

peeps,

i did some more testing of the ubitx with the homebrew specan (as opposed to the rigol), i have some good news, bad news and good news.

1. good news : the homebrew specan does a better job than the rigol.

2. bad news : the specan revealed that the even order transmit harmonics will go down below -40dbc only if we carefullly balance the bias current between the IRF510s. it is very touchy. i can't see how anyone can get it right without a spectrum analyzer

4. good news : i have replaced the two filter harmonic filters  with four filters. that also means, two more relays. it is more complicated, i know. but at least, it works. 

for those who want fewer bands (choose between 3.5 mhz and 7 mhz for the lower filter and 14 mhz and 28 mhz for the upper band), or they are alright to use external tx filters, the current design will do. for those who want all in one, wait a few days for me to update the ubitx page.

- f

Rahul Srivastava
 

Hi! Farhan..

I guess your results hold true for other 510 PP amps as well when used with other exciters as well...!!!

Rahul VU3WJM
 


On Saturday, 8 April 2017 10:29 AM, Ashhar Farhan <farhanbox@...> wrote:


peeps,

i did some more testing of the ubitx with the homebrew specan (as opposed to the rigol), i have some good news, bad news and good news.

1. good news : the homebrew specan does a better job than the rigol.

2. bad news : the specan revealed that the even order transmit harmonics will go down below -40dbc only if we carefullly balance the bias current between the IRF510s. it is very touchy. i can't see how anyone can get it right without a spectrum analyzer

4. good news : i have replaced the two filter harmonic filters  with four filters. that also means, two more relays. it is more complicated, i know. but at least, it works. 

for those who want fewer bands (choose between 3.5 mhz and 7 mhz for the lower filter and 14 mhz and 28 mhz for the upper band), or they are alright to use external tx filters, the current design will do. for those who want all in one, wait a few days for me to update the ubitx page.

- f


philip yates
 

How many bands will this cover, and does selecting 80m rather than 40m make a difference.

Still very interested.


Phil G7BZD

Jerry Gaffke
 

Building for myself, I'd try leaving the extra filters out.  Use an EFHW wire with a tuned impedance matching circuit.  Though if mass produced, it really needs an output filter at the amp.  Perhaps some holes in the board for either one band's worth of L"s and C's,  or put socket pins into those holes and use plug-in filter boards.  Can short across those holes with wires and use an external board with relays for a half dozen bands worth of filters, sell that as an option.  Not too surprising, doubt anybody in the last 70 years has sold something that relies on such balance in a push-pull amp. 

I really have to build a basic spectrum analyzer.   Perhaps just an Si5351 breakout board, a 50mhz lowpass filter, a level 7 mixer, a PX1002 86.85mhz SAW filter, an MMIC gain stage, and an AD8310 into a Nano's ADC.    Or do I need more gain stages for this to be useful?  The PJRC 3.2 ARM has a 16 bit SAR ADC, maybe add a second mixer stage to create audio in the 20 to 50 khz range and analyze that in the ARM using DSP techniques.

Jerry, KE7ER


On Fri, Apr 7, 2017 at 09:59 pm, Ashhar Farhan wrote:

1. good news : the homebrew specan does a better job than the rigol.

selfy.dtp@...
 

Farhan,

Let me first thank you for the projects and experience you share with us. I am interested to find out how uBITX behaves in terms of selectivity and sensitivity?

I also support the idea, that if you want to cover all HF band a set of (at least) four LPF for the PA is a must.

Additionally, looking at the uBITX schematic, I couldn't help but notice the bidirectional amps slightly differ form the original idea of W7ZOI and K3NHI. You used different biasing approach, but it appears the input impedance is not 50 ohms. My LTSpice simulations show 90 ohms at 10MHz and 50 ohms at 45MHz. And the "gain" transistor is biased at 15mA. Would you, please, share with us what your idea was for doing these mods.

73 de LZ1NEF

cove37
 

Have you tried using the Mitsubishi RD16HHF1?  

Thanks

Burton Ison
W5IAC




On Friday, April 7, 2017 11:59 PM, Ashhar Farhan <farhanbox@...> wrote:


peeps,

i did some more testing of the ubitx with the homebrew specan (as opposed to the rigol), i have some good news, bad news and good news.

1. good news : the homebrew specan does a better job than the rigol.

2. bad news : the specan revealed that the even order transmit harmonics will go down below -40dbc only if we carefullly balance the bias current between the IRF510s. it is very touchy. i can't see how anyone can get it right without a spectrum analyzer

4. good news : i have replaced the two filter harmonic filters  with four filters. that also means, two more relays. it is more complicated, i know. but at least, it works. 

for those who want fewer bands (choose between 3.5 mhz and 7 mhz for the lower filter and 14 mhz and 28 mhz for the upper band), or they are alright to use external tx filters, the current design will do. for those who want all in one, wait a few days for me to update the ubitx page.

- f


Ashhar Farhan
 

with the 4 filters you get 3.5, 5, 7,10,14,18,21,24.9 and 29 mhz

- f

On 8 Apr 2017 5:05 p.m., "philip yates" <phil@...> wrote:

How many bands will this cover, and does selecting 80m rather than 40m make a difference.

Still very interested.


Phil G7BZD

philip yates
 

Wow... Impressive, cannot wait for this.

Am following closely the uBitx. Better get space cleared for it.

Thanks Ashhar


Phil-G7BZD

Ashhar Farhan
 

i have measurer the input impedance of these stages to be appox 50 ohms when terminated in the 220 resistor in the output. i made these measurements with a homemade return loss bridge and the sweeperino. 
the original wes/kopski amps uses separate resistors for bias and feeback, i used just one. if you want to experiment with different feedback resistors while keeping the bias the same, you could use the original block.
the noise figure should be 14 db. i say that because i havent measured it. i do not have calibrated noise source. this figure is based on the accumulated (measured) losses of the lpf, first mixer and the first if amps figure of 6 db.
connecting the antenna brings up the noise on all bands at my qth. i must, though admit that i havent had any contacts in 28 mhz at all.
- f

On 8 Apr 2017 7:52 p.m., <selfy.dtp@...> wrote:
Farhan,

Let me first thank you for the projects and experience you share with us. I am interested to find out how uBITX behaves in terms of selectivity and sensitivity?

I also support the idea, that if you want to cover all HF band a set of (at least) four LPF for the PA is a must.

Additionally, looking at the uBITX schematic, I couldn't help but notice the bidirectional amps slightly differ form the original idea of W7ZOI and K3NHI. You used different biasing approach, but it appears the input impedance is not 50 ohms. My LTSpice simulations show 90 ohms at 10MHz and 50 ohms at 45MHz. And the "gain" transistor is biased at 15mA. Would you, please, share with us what your idea was for doing these mods.

73 de LZ1NEF

Craig Wadsworth
 

Ash,

I'm still gathering parts for my first build of your uBitX (and your Specan!) and have not even started to draw the boards, so adding filters and tweaking the code for switching bands appears trivial.

" . . . replaced the two filter harmonic filters with four filters . . . it works."

You saved us some grief and perhaps FCC fines by telling us about your tests. We grumble, we complain, and we argue, but I believe most of us are having fun and accidentally learning some good stuff along the way.

Other than encouraging fellow hams to purchase from HFSigs, let us know how we can help you.

73 de w9ctw

G1KQH
 

This was always tricky when building multiband homebrew, that's why many went for single band design, building up a bank of boxes one for each band.

Good luck Ashhar.

73 Steve


G1KQH



Jerry Gaffke
 

Off topic, just feeling my way through on this specan stuff.  

The AD8310 is spec'd for a maximum input of +/- 2 volts pk-to-pk, or 4 dBV.  So I'd want at least enough gain after my level 7 mixer to be able to drive the AD8310 to that maximum without having appreciable distortion crop up in the mixer.   That should give the instrument the 95dB dynamic range available from the AD8310.

More gain than that is a good thing, allows us to see lower level signals.  Can always use a step attenuator when looking at large signals.  But at some point the front end is generating enough noise that more gain does not buy anything.

Correct me if I've missed something here.



On Sat, Apr 8, 2017 at 06:50 am, Jerry Gaffke wrote:

really have to build a basic spectrum analyzer.   Perhaps just an Si5351 breakout board, a 50mhz lowpass filter, a level 7 mixer, a PX1002 86.85mhz SAW filter, an MMIC gain stage, and an AD8310 into a Nano's ADC.    Or do I need more gain stages for this to be useful?

selfy.dtp@...
 

Thank you Ashhar,

My 'observations' show that proper termination of the QER filter (and matching the crystals!) provides lower ripple in the pass-band together with lower losses. It really depends on what kind of crystals one is using. Mine required 150pF capacitors to ground to achieve 2.7kHz -3db pass-band and -4 dB loss @ 50 ohms at both ends.

My experience is limited only to Minima crystal filter. I modified the Rx part of the bidirectional amplifier to the original Wes/Kopski circuit with a gain of +24dB (theoretical). I am still fighting with the KISS mixer, but I guess I go into too much off topic here.

Thanks once again and 73!


On Sat, Apr 8, 2017 at 6:32 PM, Ashhar Farhan <farhanbox@...> wrote:
i have measurer the input impedance of these stages to be appox 50 ohms when terminated in the 220 resistor in the output. i made these measurements with a homemade return loss bridge and the sweeperino. 
the original wes/kopski amps uses separate resistors for bias and feeback, i used just one. if you want to experiment with different feedback resistors while keeping the bias the same, you could use the original block.
the noise figure should be 14 db. i say that because i havent measured it. i do not have calibrated noise source. this figure is based on the accumulated (measured) losses of the lpf, first mixer and the first if amps figure of 6 db.
connecting the antenna brings up the noise on all bands at my qth. i must, though admit that i havent had any contacts in 28 mhz at all.
- f

On 8 Apr 2017 7:52 p.m., <selfy.dtp@...> wrote:
Farhan,

Let me first thank you for the projects and experience you share with us. I am interested to find out how uBITX behaves in terms of selectivity and sensitivity?

I also support the idea, that if you want to cover all HF band a set of (at least) four LPF for the PA is a must.

Additionally, looking at the uBITX schematic, I couldn't help but notice the bidirectional amps slightly differ form the original idea of W7ZOI and K3NHI. You used different biasing approach, but it appears the input impedance is not 50 ohms. My LTSpice simulations show 90 ohms at 10MHz and 50 ohms at 45MHz. And the "gain" transistor is biased at 15mA. Would you, please, share with us what your idea was for doing these mods.

73 de LZ1NEF


Jerry Gaffke
 

I found the statement included below a bit puzzling.  I think what Farhan is saying is that he took one of the bidi gain blocks from the uBitx and terminated it with a 220 ohm resistor instead of the 50 ohm load for which it was designed.  The measured input impedance didn't change much from 50 ohms with the new load resistor value.

As W7ZOI and K3NHI point out here   http://www.phonestack.com/farhan/bidiamp.pdf   the bidi amps in the Bitx do not have this isolation between input and output impedance.  

LZ1NEF mentioned in his post in this thread that a simulation showed a 90 ohm input impedance for the uBitx bidi amps at 10 mhz and varying with frequency, that may be worth looking into.  Farhan's biasing approach can perform exactly the same as the W7ZOI/K3NHI biasing scheme with the proper selection of component values.  The W7ZOI/K3NHI circuit uses more parts but allows bias current to be adjusted without affecting feedback and vice-versa.


On Sat, Apr 8, 2017 at 08:32 am, Ashhar Farhan wrote:

i have measurer the input impedance of these stages to be appox 50 ohms when terminated in the 220 resistor in the output.

tArthur Bryant
 

Ashhar,

Are you going to publish the updated schematic? I have started buying parts to the uBITX, but I have not bought all of them. It would be easy for me to change to a different schematic. Unfortunately I am a electronics technician and not an engineer so it is hard for me to redesign the circuits. But I want to build a uBITX for portable us with a laptop for PSK31. Thank you for sharing your designs.

Art
KB8HGW

Ashhar Farhan
 

I am plodding through the write-up part of the new version. in the meantime, here is just the schematic. not very different.



- f

On Sat, Oct 7, 2017 at 12:17 AM, tArthur Bryant <arthur.bryant@...> wrote:
Ashhar,

Are you going to publish the updated schematic? I have started buying parts to the uBITX, but I have not bought all of them. It would be easy for me to change to a different schematic. Unfortunately I am a electronics technician and not an engineer so it is hard for me to redesign the circuits. But I want to build a uBITX for portable us with a laptop for PSK31. Thank you for sharing your designs.

Art
KB8HGW

Jerry Gaffke
 

Thanks for posting that!

Curious about your 45 mhz crystal filter.
I don't see any google hits for a 45M15F, perhaps that's a 45M15A?
The 45M15A looks tough to get from the usual distributors, but does exist.

Mouser has the  similar ECS-45K20A in stock for $6.00, that might work.
Though impedance matching network may have to change. 



On Fri, Oct 6, 2017 at 01:10 pm, Ashhar Farhan wrote:
I am plodding through the write-up part of the new version. in the meantime, here is just the schematic. not very different.
 

Ashhar Farhan
 

On Sat, Oct 7, 2017 at 2:10 AM, Jerry Gaffke via Groups.Io <jgaffke@...> wrote:
Thanks for posting that!

Curious about your 45 mhz crystal filter.
I don't see any google hits for a 45M15F, perhaps that's a 45M15A?
The 45M15A looks tough to get from the usual distributors, but does exist.

Mouser has the  similar ECS-45K20A in stock for $6.00, that might work.
Though impedance matching network may have to change. 



On Fri, Oct 6, 2017 at 01:10 pm, Ashhar Farhan wrote:
I am plodding through the write-up part of the new version. in the meantime, here is just the schematic. not very different.
 

Karl Heinz Kremer, K5KHK
 

Ashhar, back in https://groups.io/g/BITX20/message/33203?p=Created,,45M15A,50,2,0,14278989 you mentioned either the 45M15A or 45M15B - both are available on Aliexpress. One has a slightly higher insertion loss, but is a 4 pole vs. a 2 pole filter. Is there any benefit to using the "B" version vs. the "A"? 

Ashhar Farhan
 

Model A should do, we have a 12 Mhz filter further ahead of it to take care of the final bandwidth.

- f

On Sat, Oct 7, 2017 at 2:58 AM, Karl Heinz Kremer, K5KHK <khk@...> wrote:
Ashhar, back in https://groups.io/g/BITX20/message/33203?p=Created,,45M15A,50,2,0,14278989 you mentioned either the 45M15A or 45M15B - both are available on Aliexpress. One has a slightly higher insertion loss, but is a 4 pole vs. a 2 pole filter. Is there any benefit to using the "B" version vs. the "A"? 

Previous Topic Next Topic