Frequence response of through hole resistor vs power rating? (attenuators)

[coppercone2]

I have a attenuator that is busted and it calls for a 1/8 watt resistor. It looks like it was repaired before and there was not a 1/8 watt resistor in there in the first place.

I am waiting for other parts to fix this one (HP461), but I am wondering if anyone has any information on how big a difference there is for a 1/8th watt and a 1/4 watt resistor.

The resistor is 499 Ohms, so its probobly not the best thing to try to test on a VNA, and I would also need to buy it.

Are these small parts critical? The amplifier I have Fmax is 160MHz

I thought there might be a table some where that exists with this information.

I had to bend the leads out quite a bit to make it fit, it has a notch for a real 1/8th watt resistor that the previous repair man did not bother using.

FYI the attenuator had a drifted resistor, a cracked resistor that measured ok (possibly my noise source that was ruining the circuit) and a 1pF capacitor that read 500ohms.

[coppercone2]

Also does this have to do with power level or just physical size?

For instance I have a 0.4W resistors, that are some how in a 1/6th or 1/8th watt size package. Not sure how that works (way smaller then a 1/4 watt), from a real manufacturer (they advertise as reduced size package).

Are these 0.4W resistors going to be the same for RF as the 1/8th watt resistors? I have the 0.4 reduced size ones and I also have a 1/4 watt normal size ones (but the reduced size ones with the high power level are not the correct value).

Do you think I can test the reduced size 0.4W resistors vs the 1/4 watt regular resistors to make VNA plots and basically know what will happen when I get a 1/8th watt resistor in traditional size? (since if I buy a 1/8th watt resistor I will likely buy a 1/8 watt resistor rather then try to find a 'increased rating' resistor, which I just happen to have).


Is there some kind of other magic going on with the reduced size higher power ones? I don't know how they make it like 3x smaller then a 1/4 watt but then it has 40% more power rating).

[mag_therm]

Hi Cc,
Do you have enough  info about , or by tracing circuit of, the attenuator  to be able to calculate the power loss in that resistor at max output?

[coppercone2]

the power loss is well handled by the 1/8th resistor, its a LNA with 50 ohm input impedance, the input voltage limit is low, the maximum output power is ~7dBm, so the input should not really get above -30dBm since it has 40dB gain (and the attenuator is front end and cuts input in half, it does not change gain)

its a LNA input attenuator not output attenuator

so the non attenuated maximum input is -30dbm to get the 7dbm output power maximum, if you swap in the 20dbm attenuator it should be -10dbm input maximum, since -10dbm goes to -30dbm and that is gained to ~10dbm, rounding the numbers hard. that should be 100uW input for -10dbm I think.

the manual specifies that maximum allowed input is 1Vrms or 2V pulse but the output max is 0.5Vrms so yeah

https://xdevs.com/doc/HP_Agilent_Keysight/HP%20461A%20462A%20Operating%20&%20Service.pdf

At 1Vrms input the dissipation in a 600 ohm resistor is like 1mW, 3.5mW for the DC signal of +2V if you figure that the attenuator resistor and the load resistor are in series to divide 2/550

[mag_therm]

Then rating isn't critical,  you could just try for a physical size close to the original, so leads are approx same.
That is about all you can do, isn't it?

[coppercone2]

Well I mean if someone thinks that its a grievous mistake I can order the proper part while I am waiting for others, thought it might be interesting to hear about what kind of performance people have got from through hole by reducing size

[joeqsmith]

Maybe you will find something in this thread of use:
https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/

[coppercone2]

Hmm David Hess says that they are a "little better" and that 1/4 watt in that configuration is good to like a GHz. This is 160MHz. I guess it should be fine.

But I think I will order the 1/8 watt resistor and replace some of other resistors on the PCB. I began messing with the amp before I realized the attenuator was fucked up, so I replaced some resistors with 2 resistors in parallel of dubious quality.. might as well replace all those with proper parts if I am gonna do the order and keep it at HP quality

finally finished my e96 1/4 watt resistor by decade, but it looks like the power levels are all wrong to make the collection work as it should

[T3sl4co1l]

Generally speaking, resistors exhibit stray capacitance for larger values (over 100R or so), or inductance for smaller.  The amount is given by body size, so the bandwidth is maximum for values near this characteristic impedance, and proportionally worse for values distant to it.

So, if a 220R 1/8W resistor goes capacitive around maybe 2GHz, expect 2.2k to go similarly around 200MHz, etc.

There's a small excess inductance due to circuitous (usually spiral) traces in resistor construction, but this is a small contributor for carbon film resistors; it's mainly noticeable for wirewound types.  Likewise, excess capacitance due to body materials (usually alumina substrate, has a fair κ), or heatsinking or housings (beware with thin-film TO-220 etc. types, also metal-case wirewounds).

Who knows what the reference value is for a given part family, you may have to look it up or test.  But it will be fairly consistent within that family.

Tim

[coppercone2]

Wow the attenuator mechanism on this thing is a joke. There is a bent piece of brass that interacts with a rounded bar of aluminum fixed to a wafer switch. So they are making a thin square interact with a crappy high wear aluminum piece. It must be sensitive to fractional degrees, god forbid someone put a curved surface on there. Looks like that amplifier has gone missing.

Have not been this disapointed with HP construction method since I ran across that 100V "precision" power supply with its plastic switch matrix, that was built like a garage door with literally a bunch of panel meters and shit sandwiched together like a garage door on the front panel with no screws. This old HP stuff is really hit or miss on quality.

I am defiantly not looking to align shitty under built cams with precision sanding in this hobby.

If you have one of these, I recommend sanding the cam down and JB welding a piece of brass shim around the actuator to make some kind of decent surface so its not aluminum. Maybe that would work but I lost my patience on this thing. I think they must have transfered the designer of this attenuator to their printer department, because its about as fun to work on as a printer. This is the literal mechanism that is used to switch in the attenuator :  .


it looks like a scorpion too.


you know what, that whole attenuator is insane. Why don't you make a module that has a soldered BNC connector that screws in to the front of the unit. Instead you have some cockamy bullshit assembly that has a BNC Connector that is press fit into a piece of a switch (seriously, they have the BNC center conductor solder cup jam into a bent piece of metal) and fixed with a set screw. What the hell is the benefit. You need more screws and more shit to fit together to do it that way. And change the switch to the other kind (normally open/closed) so that the attenuator switch is simply depressed by maximum knob rotation rather then have it necessary to have the attenuator work on the middle switch setting (i.e. why the fuck does the cam have to engage the switch from both directions?). I can't figure out what they were thinking. IMO its nuts to sell this mechanism. If you just press down on it when the switch is fully turned, that would be fine, it would just press the switch on the top like its meant to be. you might not even need a god damn spring if you did it that way.   they treat the switch assembly like a punching bag not a depression switch the way it is. I totally hate this amplifier. Its a push button switch not a 'lick button' switch. god damn what a stupid design, and also the grounding path sucks really bad and goes through so many screws and aluminum pieces. . This could have been so simple.

[Hamelec]

if you are pissed you should change your Hobby

[coppercone2]

why, its just a shitty old design. You come across some serious trash every decade.

What I will do is throw away the PCB and put a new nice amplifier inside the box with a real attenuator.

Pressing a solder cup into a bent piece of metal and fixing a BNC into an aluminum ring with a set screw is trash. Totally appalled by the construction quality of this amplifier. Made in the USSR I assume. 

HP got away from this garbage since the 3403C thermal RMS meter, they finally decided a panel cutout going into a can with a properly mounted BNC on it is superior to trying to make thread on VHF electronics based on something that might make sense for a small motor. The design in the HP461A attenuator looks like something someone came up with when they crashed a motor cycle into a dumpster. The best I can figure other then insanity is that sales absolutely demanded that there cannot be a screw on the front panel, and a small sheet metal bracket that mounts it to the chassis properly is too expensive.

Thank your lucky stars someone came up with the idea of a RF can with screws on top that can attach to something with screw holes on the bottom. At least this should only effect the 461 and 462 amplifiers, the 465 is fine. If they put a side ways wafer switch that turns to depress the button on top of the amplifier it would be better too, even if the attenuator is a sad sad construct.


This is literally: drive a nail into the outlet and wedge a coke can under neath

[coppercone2]

In case my point is hard to understand...

Mechanism in use by HP461A:


proposed mechanism


I heard the 461A attenuator switch inspired this (original HP designers appeared in this scene wearing suits):

[coppercone2]

I think I might have a go at removing the aluminum decapitator from the waffle switch and to mill a slot on the front panel for a lever switch that properly pushes down the attenuator. It might be a good project for the MF70 mill to make a quality mechanism out of brass (I call it the finger)

Possibly the first useful clock work construct in my lab. I feel a little bad throwing away all those coils in there... but that switch is not acceptable by any means. Maybe I will make a collet on the clicky switch to accept the front panel BNC connector as well.

[coppercone2]

I decided to take another crack at it, I wound a spring by drilling a hole in a old damaged eyeglasses repair screwdriver with a carbide drill bit and a proxxon mini sensitive drill press, then I wound a spring around it that I tore out of a broken pen. I used nylon pliers to wind it, and carbide flush cutters to cut it.

I made the new spring lever arm thing out of some kind of alan bradley spare part for something or another for furnaces (piece of spring steel with same dimensions).

I used a oil stone and other stones to smooth out the spring lever, and I plan on giving the guilotine a bit of sanding. too lazy to try to make an alternative mechanism, and I felt bad throwing the amp away

I just need to spot weld a section of screw driver to the lever arm to make a mounting stud for the spring. but after i sanded it down with the diamond file on a leatherman, it does not appear to want to move. I feel like locating the stud is going to be a pain. maybe a drop of epoxy. if I do anything but the natural position its going to want to get jammed.

[coppercone2]

lol a bunch of pads came off the board (the coaxial to board connection on this PCB is ridiculous), so I drilled a hole and put a big copper eyelet in there to bond the parts at a center point. and the indicator light fell apart so I ordered a handful of replacement s that should fit.

I swear if HP put like a handful of turret terminals on this PCB it would go from frustrating bull shit to top notch 

[coppercone2]

also got the test clip fixture for the IA.

Its the SMD type, but I have a feeling you can easily test leaded components on it.
https://www.keysight.com/content/dam/keysight/en/img/migrated/cmc_upload/ck/WN/PLWN-Kobe/16092A_support_image3.png

Does anyone know if there is a problem with say using a small plastic clamp to attach the low side of the component to the ground plane? maybe through a thin shim of gold so it does not scratch the plate?

[coppercone2]

ressurected the 461A. Works. Has IEC input power and a fixed attenuator. bit like a car muffler.

Messed with the gain resistors, it gets between 39db to 40.5db gain between 1khz and 1MHz on a 34401A.

Was chasing a ghost for a little while because I forgot the DMM is no load (no load vs 50 ohm), for the gain testing. where did 10db go?! into the 50 ohm input resistor!