Manhattan island cutter and HF pre amp

[Malcolm T]

Just thought a few images might be worth posting up of two cutters i fabricated up .
I decided to go Manhattan style as i had read about some nice results of this construction method, good for experimentation and as a final build style. Problem was the cutter material and fabing a cutter of 7mm dia or less. the four tooth cutter is from a mild steel tube, filed out with small modelers files, it didn't stand up to the glass fibre board though .

The second attempt was harder to make but has superior resilience to the glass fibre, made from a 6.5mm shank of a drill bit i hacked off with a cutting disc on an electric grinder.
The cutting tip was first made by changing to a 120 grinding disc and grinding a slope across the end of the shank then slowly and carefully grinding away excess either side of the high side to leave a small tip of about 1.4 mm proud, not easy with a hand held mains grinder.
This stands up very well against the fibre glass.

The pre amp consists of two BC547, rescued from an old TV from the dump, as were the trimmers, all tested and appear to be ok. But im not convinced of the circuit design used or the 0.047uf caps, and the trimmers but i thought i would give it a go, i got mixed results from it, so am in the process of rebuilding to a more conventional layout . Any comments on the pre amp circuit will be appreciated.

[AdrianH]

I like the idea of the cutter, wonder if an 8mm dia silver steel bar could be drilled 7mm and then hardened after filing the cutting slots. Possibly could do that in my lathe if I had coolant to keep the silver steel from getting hot whilst drilling.
I typically solder double sided PCB board on one surface of another to form the islands.

Adrian

[Radio Wrangler]

If you're building single sided, you don't even need the cutter, just gut (guillotine, saw or whatever) some raw board to make little squares from raw board. Glue them on wherever you want with cyanoacrylate adhesive.

Mount ICs upside down on cyanoacrylate and the legs act as stand-offs. Similarly decoupling capacitors become stand-offs.

As an added bonus, your groundplane continues under all pads, so its impedance isn't spoiled by cutouts.

THere are many ways of breadboarding, no two people do it the same. I've developed ultra-low noise RF stuff into the GHz region.

David

[Malcolm T]

Yes many ways to achieve a result , interesting results i had . I wanted to make the cutter from a 6mm shank but i never had one to hand , it was a bit fiddly with the grinder, the board was held in position by two roll pins pressed into a wood block and then held down with g clamps and wooden clamping fingers like on a milling machine , an absolute must. The pillar drill was a bit fiddly to set with the adjustment setting screws, really just need a fine cut through the copper and that's enough.
Theres loads of room here for experimentation , the glue technique sound very interesting too.
Ghz region sounds interesting ,

[G0HZU_JMR]

The pre amp consists of two BC547, rescued from an old TV from the dump, as were the trimmers, all tested and appear to be ok. But im not convinced of the circuit design used or the 0.047uf caps, and the trimmers but i thought i would give it a go, i got mixed results from it, so am in the process of rebuilding to a more conventional layout . Any comments on the pre amp circuit will be appreciated.

Hi Malcolm, I had a look at your preamp circuit and this is a very familiar circuit. This type of amplifier has high reverse isolation and quite well defined port impedances and the impedance is defined by the series input and output resistors. I tried to work out the bias resistor values from the photo and I think they are 100R, 680R and 820R but I can't be certain. If so it looks like the amp runs at a fairly low bias current in both devices and this will mean this amplifier will generate significant distortion terms if connected to a large HF antenna at night.

This type of amplifier has negative feedback to help define the input port impedance and this can also help with distortion cancellation. However, the phase of the negative feedback needs to be well defined in order to get optimum results. Normally, high Ft BJTs are used in circuits like this to help maintain a well defined input port impedance and also to help to minimise distortion. the high Ft BJTs will have lower internal capacitances and this will help define the feedback better over a much wider bandwidth. BC547 transistors will be adequate but not ideal here I think.

It looks like the bias current in both devices is in the region of 6mA and so this will mean this amplifier will be quite weedy in the presence of large signals. Also, when even larger signals are fed into this type of amplifier the feedback can't function correctly any longer and the distortion levels will rise much faster than expected.

This type of amp is best suited to scenarios where the input signal level is well controlled and is unlikely to overload the amplifier. The other niggle with these amplifiers is that the noise figure will be quite high. I'd expect it to be over 11dB for your circuit based on prior experience. One way to improve the noise figure would be to fit a 100uH RF choke in series with the (100R?) shunt resistor at the base of the input transistor. This usually improves the noise figure by several dB. However, this improvement in noise figure probably wouldn't be noticeable on the noisy HF bands.

[G0HZU_JMR]

I think your circuit would have better signal handling if the bias current was improved for both of the BC547 BJTs. By the way of a comparison I still have an old version of this circuit here that was built using BFR91 (very high Ft) transistors and I run these devices at a higher current.

This test amp has flat gain to several hundred MHz and the input and output VSWR is better than 1.1:1 across the HF band. The reverse isolation is better than 65dB across the HF band and the noise figure is less than 6dB with a 100uH choke fitted. I used different bias resistors to run the BFR91s at a higher current and this improves the large signal handling.

It is essentially the same circuit as yours but with faster BJT devices and a higher bias current in both BJTs. I use this test amp whenever I need high reverse isolation and low VSWR on the input and output ports up into the VHF region. The high reverse isolation sets it apart from regular MMIC based 'gain block' amplifiers but it only works well up to a few hundred MHz.

[joebog1]

Isnt that interesting!! ?.
I was just going through my junkbox transistors and look what I found. You can tell by the patina that these are brand new and origional 1978 editions.
They will be in your mailbox soon Malcolm.

Joe

[G0HZU_JMR]

Wow, those BFR91 devices are quite rare if they are 1978 vintage. I've still got a bag of them here that are probably about 30 years old.

I do have my own s-parameter models for the BC547B but the nearest bias point I have is 5Vce and 5mA Ic. This is a bit on the low side in terms of bias current but see below for a quick and dirty simulation up to 50MHz. This is a small signal AC model of the BC547B so I don't need to include a power supply. The BC547B was biased at 5V and 5mA when I extracted the model.

This shows that the port VSWR slowly degrades with increasing frequency. The gain response is probably fine for HF use and the reverse isolation is very good below 10MHz. By 30MHz it degrades to about 35dB. The models were taken of a 'Multicomp' BC547B purchased from Farnell and maybe I should have measured a device from a known manufacturer.

The BFR91 is probably a bit OTT for use only on the HF bands. The board layout would have to be quite tight and smaller components would be needed. Otherwise the amp could have stability issues up in the GHz region. I can post up images of my BFR91 test amplifier if this would be of interest. It really should be in a screened enclosure but it seems to work fine as it is whenever I use it.

[joebog1]

The date code would indicate 1978 Jeremy. I will never use them, so Malcolm can play with them.
Im still searching out transistors so we will see what I can find.

Joe

[Malcolm T]

Thanks for the reply s. Well I,m no circuit designer Jeremy so I appreciate your explanation here, strange how the most basic circuit can be so complex fascinating stuff !
I,m going to re-build with a conventional resistor bias network for both stages and a stage coupling capacitor. I have seen several circuits for pre-amps but thought I would build this very basic one and see what it could do.
The corrosion on the transistor leads should come off quite easy Joe , thanks for those.
So I have sent myself back to learning again with basic transistor theory on biasing etc LOL.
How do you set up your base biasing Jeremy , do you use the published device curves for the device or just through your experience , how do you calculate your particular required base current then ?. Because as I understand it the starting point for all calcs is the base current , right or wrong ?.
1978 , was it that long ago Joe !!!.

[kellys_eye]

I've successfully used a paper hole punch and 0.5mm copper clad material to make 'hundreds' of circular lands that can be adhered to a ground plane.

The punched material ends up with a slight curve due to the puch action but stacking a few and giving them a squeeze with pliers makes them all flat again.

[G0HZU_JMR]

How do you set up your base biasing Jeremy , do you use the published device curves for the device or just through your experience , how do you calculate your particular required base current then ?. Because as I understand it the starting point for all calcs is the base current , right or wrong ?.

In this case the RF and DC feedback both use the same resistors and so the circuit has to be designed carefully if you want to achieve a target RF power gain at a certain DC bias point for each transistor.

Your circuit is a special case where the emitter resistor R4 in the diagram below is missing. Normally this resistor is included if the designer wants to have control over the DC operating point and also the RF power gain of the amplifier.

I put together a quick excel spreadsheet that offers a fairly crude attempt at calculating the resistor values for a particular power gain and for a desired collector current in TR2. The spreadsheet assumes that TR1 will always be biased such that the collector voltage at TR1 is at half of the supply voltage. The spreadsheet allows for the emitter resistor R4 and there are entries for power gain and the desired TR2 collector current.

Have a look at the spreadsheet below, if you want to have a play with it I think excel spreadsheets can still be attached to a forum post. I haven't tried since the recent forum upgrade.

[joebog1]

1978 , was it that long ago Joe !!!.

Thats what the date code says on the transistors !!

Joe

[G0HZU_JMR]

For use across HF I think a pair of 2N3904 BJTs should perform better than the BC547. The 2N3904 (assuming it is a genuine 2N3904) is sourced from process 23 and this manufacturing process is much better suited to a fairly demanding RF circuit like this. The BC547 is normally sourced from process 04 and transistors from this classic old process group aren't a good choice for circuits like this.

You can see in my output VSWR plot (dark blue trace) in post #8 that the output VSWR degrades badly towards 30MHz when using the BC547. Normally, the impedance at the emitter should be very, very low when there is about 6mA emitter current and so the output VSWR should be defined by the series 47R resistor at the output. It looks like the BC547 is misbehaving quite badly here up at 30MHz. I'm tempted to actually build the circuit to see if it really is this poor. I simulated the same circuit using a 2N3904 and it was much better. Not as good as the BFR91 but the 2N3904 is probably a good compromise as the risk of instability will be much lower with the 2N3904.

[Malcolm T]

This is getting interesting now , thanks for the spread sheet . Very strange not having a conventional bias for the TRs , never seen a circuit like it before . So now im into transistor basics and thinking about a design program for Linux , not Lt spice ! just to get a better understanding of where to pitch the biasing for the circuit and attempt to obtain acceptable results and compare transistor types .
Any ideas on a circuit design program for RF anyone ? .

[David G4EBT]

I'm not a huge fan of 'Manhattan style' construction, though it's often used for small experimental projects and is much favoured in QRP circles, often for quite complex projects with impressive results.

The 'QRPme' website 'across the pond' sells sheets of little square 'tiles' referred to (for reasons that elude me) as 'Limerick Construction'.

Quote:

BIG MeSQUARES panels have 225 square Limerick construction pads which are easily snapped apart and then glued down. Each pad is .25 inches square, pre-tinned and has a solder mask and silk screen outline around each pad. No metal punches required!

End quote.

Neat idea.

At $10 US (£7.60 GBP currently), the little squares work out at 3.3 pence each in the UK.

But with P&P to the UK, it bumps the price up to $21 US (£16 GBP), so each tile would cost 7 pence in the UK. (Unless of course several boards at a time were purchased).

I guess it would be just as easy, and much cheaper, to cut strips of laminate say 6mm wide, then cut the strip into tiles at 6mm intervals.

http://www.qrpme.com/?p=product&id=MESb

There are also QRP 'Proto' Boards, which look quite a snazzy idea:

https://miscdotgeek.com/a-new-protot...omebrew-radio/

Anything which help to keep homebrew projects and 'QRP' alive - (the last refuge of technical, experimentational and constructional amateur radio), has to be a good thing. Long may it continue.

[G0HZU_JMR]

Here's the actual excel spreadsheet as an attachment.

This amplifier is fairly straightforward in terms of how it operates. There is a high gain inverting amplifier TR1 and this has oodles of voltage gain on its own. Much, MUCH more gain than is required in the finished and complete amplifier design. However, TR1 is followed by an emitter follower TR2 and this has a very low impedance output at the emitter pin and this inverted (anti-phase) signal is fed back to the input at TR1 base (via R5) to manage the overall amplifier gain to a more controlled and sensible level.

This anti-phase feedback current via R5 will try and cancel/null the RF input drive current fed in via R1 and this will reduce the effective gain of TR1 such that it has the desired gain dictated by the excel spreadsheet.

In this sense this amplifier is a bit like an inverting op amp as there is loads of open loop gain in TR1 and a very low impedance node at the output (TR2 emitter). When the feedback is connected via R5 there is also a kind of virtual ground at the base of TR1 where the input drive signal current into TR1 base via R1 is cancelled by the anti-phase current fed back via R5. The amplifier will therefore reach equilibrium when the voltage gain from C1 to the emitter of TR2 = R5/R1. This is very similar to an inverting op amp in this respect.

Of course, all of this behaviour begins to fall apart as the frequency is increased as TR1 and TR2 are not perfect transistors. The internal resistance and capacitance of the BC547 will cause undesirable phase shifting and level shifting and the quality of the virtual ground at TR1 base and the low impedance at TR2 emitter will degrade with increasing frequency. This will degrade the gain flatness and also the port VSWR as the frequency is increased. Higher Ft BJTs will behave much better here up into VHF.

Resistor R6 is needed for the DC biasing but it also acts as an undesirable place to waste RF energy as some of the input RF current and the feedback current is shared here and wasted in R6. This is why the noise figure will improve with a 100uH RF choke fitted in series with R6. This won't affect the DC biasing but the RF choke prevents R6 from absorbing RF signal energy.

[G0HZU_JMR]

See below for a quick design demo for a 14dB gain amplifier using 2N3904 BJTs.

I've posted up a screenshot of the excel spreadsheet showing the final design and there is also a screenshot of a simulation. The simulation shows about a x10 voltage gain from TP1 to TP4.

Because there is a series 47R resistor at the output (to define the 50R port impedance) there will be a divide by two in terms of voltage here. So a design that asks for a 14dB gain will have about a x10 voltage gain from TP1 to TP4.
You can also see that at TP2 (TR1 base) there is very little signal voltage as this node acts as a virtual ground because of the negative feedback via R5. So there should be very little RF voltage seen here. There's a spectrum plot to show distortion although the 2N3904 model in Genesys isn't that accurate in this respect.

You can see the excel spreadsheet managed to produce the correct 14dB gain in the simulated amplifier and it also managed to bias both transistors as expected with about 8.8mA in TR1 and 15mA in TR2 and TR1 collector is at 6V which is half of the supply voltage.

The 50R input and output port VSWR is very good using the 2N3904 according to the simulator. It probably won't be quite this good with real devices but it will still be quite good.

[G6Tanuki]

It all reminds me of 'Blob Boards' which were sold in the 70s and 80s to the hobbyist

[Terry_VK5TM]

Just adding my thanks to Jeremy for the info he has added - I played with this circuit some time ago and didn't have much luck with it.

I do note seeing that circuit some time ago that split the R5 feedback resistor with a cap to ground in the middle to tailor the frequency response.

Might give it another go now with the extra info above