HP35s any good for EE?

[spaff]

Firstly, hello!

Secondly, I've been recommended an HP RPN calculator by a friend. After doing some research, it looks like the HP35s is good enough for what I need. I'm currently using a Casio FX991MS and TBH it sucks, is worn out, has no persistent memory and the matrix implementation is pretty poor. I tried Casio's mid-range 9750G but all I can say is "meh" and I borrowed an HP50g but it's too easy to get lost in that and you can't hold the entire model of the machine in your head without years of work with it. Also borrowed a TI nSpire and I'd rather gouge my eyes out than use it again - everything is painful. TI 89 has been written off as too complicated plus I don't need a CAS (just mech pencil and a brain). Linear regression would be useful. I also don't need graphing ability so would rather something a bit smaller than a full graphic calculator. PC eliminated as well as input is clumsy.

I do a lot of analogue EE work and tend to use quite complex functionality i.e. matrix determinants, solver, numeric integration, complex numbers (polar/rect and converting between them) and heavy use of engineering notation.

I did read somewhere about some problems with them but I'm wondering if that's going to be an issue for my use cases or if they have been resolved since.

Is this going to be suitable, reliable and reasonably accurate in your professional opinions?

[rprosperi]

(04-15-2015 03:52 PM)spaff Wrote:  Firstly, hello!

Secondly, I've been recommended an HP RPN calculator by a friend. After doing some research, it looks like the HP35s is good enough for what I need. I'm currently using a Casio FX991MS and TBH it sucks, is worn out, has no persistent memory and the matrix implementation is pretty poor. I tried Casio's mid-range 9750G but all I can say is "meh" and I borrowed an HP50g but it's too easy to get lost in that and you can't hold the entire model of the machine in your head without years of work with it. Also borrowed a TI nSpire and I'd rather gouge my eyes out than use it again - everything is painful. TI 89 has been written off as too complicated plus I don't need a CAS (just mech pencil and a brain). Linear regression would be useful. I also don't need graphing ability so would rather something a bit smaller than a full graphic calculator. PC eliminated as well as input is clumsy.

I do a lot of analogue EE work and tend to use quite complex functionality i.e. matrix determinants, solver, numeric integration, complex numbers (polar/rect and converting between them) and heavy use of engineering notation.

I did read somewhere about some problems with them but I'm wondering if that's going to be an issue for my use cases or if they have been resolved since.

Is this going to be suitable, reliable and reasonably accurate in your professional opinions?

You didn't mention if the calculator you are seeking must be new or cheap. If the answers to those are not necessarily yes, then look closely at an HP-42S, available used on common auction sites, or possibly through posting a WTB on in the Classified Fourm here.

The 42S does everything you mentioned far better than the 35S IMHO, is easier to learn and use, is smaller, probably as fast or faster (?), also has 2-line display, and there are probably thousands of programs of all types published for it here and in other sites. The 42S will also run most 41C programs unchanged.

A new 35S is $59 IIRC, and a nice used 42S will be at least $200. In my opinion, there is no comparison; get the 42S.

[Tugdual]

Did you consider a prime? Otherwise I would prefer a 50g. The 35s is very limited regarding complex numbers and the 50g does a good job. I agree the 50g is so capable that you can easily get lost but you may also customize it for your specific needs. Prime is good but I find the 50g more versatile and less focused on just school studies.

[pito]

(04-15-2015 03:52 PM)spaff Wrote:  I do a lot of analogue EE work and tend to use quite complex functionality i.e. matrix determinants, solver, numeric integration, complex numbers (polar/rect and converting between them) and heavy use of engineering notation.

During my EE studies I had a teacher who was using HP25 and he was able to solve almost any RLC network within minutes Maybe a tablet with Spice simulator on it will do better when messing with analog..

[spaff]

Thanks for the replies.

Cheap is a bonus here as I'm not exactly rolling in cash. I just had to blow most of my spare cash on a new oscilloscope unfortunately. I'd rather grab something new and replaceable than an older model here.

I just had a play with the prime emulator but I found the interface a bit counterintuitive and it didn't feel finished as a product. Might be right with the 50g yet - I'll borrow it again and do some more playing.

Looked at tablets and CAS software but I really like to escape into another dimension where there are no distractions other than the smell of burned fingers on my soldering iron. I'm slightly worried that the 50g would be another distraction.

[pito]

If you work with analog electronics seriously then sooner or later you will do with complex circuitry with active components as well, where you will need a simulator like LTSpice and friends (with active component's models).. No chance to do it with a calculator..

[spaff]

I use LTspice occasionally as a thinking tool but things rarely work to such ideal models.

If you use abstractions and build simple models then you can do anything without complex tools, hence why we deal with ohms law instead of Maxwell's equations.

[emece67]

I've used for EE many HP calcs, from my first ever HP, a 28C, to my last, an 48GX, with 15C, 41, 32S, and 42S in between. From a pure practical perspective, IMHO, the sweet point for everyday EE use is between a "top-end" RPN machine (say, the 42S) and a "low-end" RPL one (say a 28S or 48S/SX).

Being the 42S scarce and expensive and the 28S also scarce, I think that a used 48S/SX, that you can buy at really good prices, is a nice option for EE work. It transparently works with complex numbers and matrices and, as a premium, it also works with units, a capability I really miss in RPN calculators. The 48, being it one of the first incarnations of the RPL style, lacks some later "features" I think are not necessary for an everyday calculator (say: big CAS, too many "menu applications"*, equation libraries,...). It does also have the classic HP keys on its classical orange/blue colour theme.

Of course it is not perfect, I miss better probability distribution support & eigenvalues (48G does this last).

Hope this helps, regards.


(*) I mean, such things as the menus you can find in the 48G to set system flags, or entering the solver, the 48S lacks. I think they turn the machine slow and inefficient to work with.

[ElectroDuende]

(04-17-2015 12:33 AM)emece67 Wrote:  I've used for EE many HP calcs, from my first ever HP, a 28C, to my last, an 48GX, with 15C, 41, 32S, and 42S in between. From a pure practical perspective, IMHO, the sweet point for everyday EE use is between a "top-end" RPN machine (say, the 42S) and a "low-end" RPL one (say a 28S or 48S/SX).

I don't have such a long history using HP's for electrical engineering; just passed from a Casio FX-4200p to a 48G.

(04-17-2015 12:33 AM)emece67 Wrote:  Being the 42S scarce and expensive and the 28S also scarce, I think that a used 48S/SX, that you can buy at really good prices, is a nice option for EE work. It transparently works with complex numbers and matrices and, as a premium, it also works with units, a capability I really miss in RPN calculators. The 48, being it one of the first incarnations of the RPL style, lacks some later "features" I think are not necessary for an everyday calculator (say: big CAS, too many "menu applications"*, equation libraries,...). It does also have the classic HP keys on its classical orange/blue colour theme.

Of course it is not perfect, I miss better probability distribution support & eigenvalues (48G does this last).

I totally agree. Well, I'd recommend a 48G+ that will have the "black LCD" that is a dramatic improvement over 48S and 48G blue ones.

For an Electrical Engineer the use of complex numbers is crucial, and you will need a calculator that can work with them as they were reals. The 48 series does that; I'm not so sure about the 35.

(04-17-2015 12:33 AM)emece67 Wrote:  (*) I mean, such things as the menus you can find in the 48G to set system flags, or entering the solver, the 48S lacks. I think they turn the machine slow and inefficient to work with.

But you can still use the "old way" to set flags or use the solver...

[walter b]

(04-17-2015 12:33 AM)emece67 Wrote:  ... it also works with units, a capability I really miss in RPN calculators.

Please, would you mind enlightening me why you need that?? TIA.

Background: On our side of the Atlantic Ocean, people use SI units. Doing so, you can throw in any such units (V, A, Ω, F, T, Cb, ...) in an equation and the result will be returned in the corresponding correct unit - as long as the equation is correct, of course.

d:-?

[emece67]

(04-17-2015 09:09 AM)walter b Wrote:  

(04-17-2015 12:33 AM)emece67 Wrote:  ... it also works with units, a capability I really miss in RPN calculators.

Please, would you mind enlightening me why you need that?? TIA.

Background: On our side of the Atlantic Ocean, people use SI units. Doing so, you can throw in any such units (V, A, Ω, F, T, Cb, ...) in an equation and the result will be returned in the corresponding correct unit - as long as the equation is correct, of course.

d:-?

Cause I work, many times, with old documents that use non-consistent unit systems.

Also, because I rely on unit consistency to check the validity of some calculations. e.g. Working with electro-mechano-acoustical systems, you have 3 kinds of Ohms ("classical" ohms, mechanical ones and acoustic ones), and 3 kinds of "currents" and 3 kinds of "voltages"; working with a big number of different magnitudes is easy to forget that you need to multiply/divide by some other magnitude to get the result you want (specifically, when you cross some "boundary" between the 3 domains: electrical-mechanical-acoustic). With an unit-aware calculator, if, after a long calculation with many different magnitudes, I get a force in N/m2, I instantly see that I've forgot to multiply by a surface, so I re-check my calculations. Note that this is a problem even if you are using the SI (as the SI does not make you error-free).

Also 2: because many modern (and old) documents do not clearly state the magnitudes they are measuring. E.g. some authors state the power loss in a magnetic core in W, but others in W/m3 and still others in W/m2, and the remaining ones in W/kg and, even more, some use J instead of W. But they do not clearly say what definition of "magnetic loss" they are using. Blindly using their equations without checking the units is a safe recipe to non working systems. As above, if you get the current across a coil in, say, A/sqrt(s·Kg), you know you have an error, somewhere, involving a mass and a time; perhaps you and the author you are consulting use different definitions for some magnitude involving a mass and a time. This is a problem even if you adhere to the SI (as the SI does not state what is the definition of magnetic loss in a core).

Also 3: because sometimes you are not calculating an equation you see in a book or paper, but you are creating such an equation and, of course, you cannot be absolutely sure that you have not committed an error. Checking unit consistency in such equation is a good method to be on the safe side. You can do the check manually but, well, I have a little machine that can do it for me!

Also 4: because in the previous cases (perhaps with the exception of Also 3), the approach of checking all unit previous to the calculations and, if any problem is detected, converting all magnitudes to the SI is, IMHO, not very practical.

Regards.

p.s. I think I'm at the very same side of the ocean that you are.

p.s.2. I also miss this unit-aware behaviour in other tools, such as Matlab. The Mars Climate Orbiter wouldn't have crashed using such tools (although the real problem here was some people did not adhere to the SI).

[emece67]

(04-17-2015 08:15 AM)ElectroDuende Wrote:  

(04-17-2015 12:33 AM)emece67 Wrote:  (*) I mean, such things as the menus you can find in the 48G to set system flags, or entering the solver, the 48S lacks. I think they turn the machine slow and inefficient to work with.

But you can still use the "old way" to set flags or use the solver...

Of course, it's simply a matter of taste. I find then distracting and, thus, is some way disgusting. For me they are out of the absolute consistency and uniformity of RPL.

[emece67]

At the risk of being OT.

Another thing I miss in calculators is an ENG mode working like this:

Usual ENG ----- "alpha" ENG
123.456e12 ---- 123T456
123.456e9 ----- 123G456
123.456e6 ----- 123M456
123.456e3 ----- 123k456
123.456e0 ----- 123.456
123.456e-3 ---- 123m456
123.456e-6 ---- 123u456
123.456e-9 ---- 123n456
123.456e-12 --- 123p456
123.456e-15 --- 123f456 (not a!!)

Does anybody know of a calculator working this way?

[ElectroDuende]

(04-17-2015 11:50 AM)emece67 Wrote:  At the risk of being OT.

Another thing I miss in calculators is an ENG mode working like this:

Usual ENG ----- "alpha" ENG
123.456e12 ---- 123T456
123.456e9 ----- 123G456
123.456e6 ----- 123M456
123.456e3 ----- 123k456
123.456e0 ----- 123.456
123.456e-3 ---- 123m456
123.456e-6 ---- 123u456
123.456e-9 ---- 123n456
123.456e-12 --- 123p456
123.456e-15 --- 123a456

Does anybody know of a calculator working this way?

Mi old CASIO FX4200p does almost that!! (just that 1e-15 is f) ;-)

I think will represent numbers like 123.456k no 123k456 as you propose. (i hoe I remember to check it later when back home).

[spaff]

All mid to high end Casios have that feature now (9750G and 9860G). It's been around for quite some time, at least since 1996

It's hidden in the option menu.

Emulator: http://oud.digischool.nl/wi/rekenmachine...lator.html

To reproduce:

1. Hit 1 to go into immediate entry mode.
2. Shift-Menu (setup)
3. Scroll down to "display"
4. Press F4 to use engineering symbols
5. Exit setup
6. enter value 123.45e-6 and press enter ... shows as 123.45 mu.
7. Press option, F6, F6, F1 (engineering symbols menu)
8. Gives you numeric engineering suffixes to use after numbers.

Not a great implementation but it makes life quite easy. That and FIX 3 and it's a half decent calculator.

[emece67]

(04-17-2015 12:20 PM)ElectroDuende Wrote:  (just that 1e-15 is f) ;-)

Oops!

(04-17-2015 12:53 PM)spaff Wrote:  All mid to high end Casios have that feature now (9750G and 9860G). It's been around for quite some time, at least since 1996

It's hidden in the option menu.

...

Just tested the emulator. I didn't have a clue that Casios are working this way. Nice and, as you say, making life easier.

[walter b]

(04-17-2015 12:20 PM)ElectroDuende Wrote:  Mi old CASIO FX4200p does almost that!! (just that 1e-15 is f) ;-)

... and 1e-18 is a.

ElectroDuende Wrote:I think will represent numbers like 123.456k no 123k456 as you propose.

I hope so.

(04-17-2015 12:53 PM)spaff Wrote:  Not a great implementation but it makes life quite easy. That and FIX 3 and it's a half decent calculator.

FIX 3? I hope ENG 3.

d:-)

[spaff]

(04-17-2015 02:26 PM)walter b Wrote:  FIX 3? I hope ENG 3.

d:-)

Fix 3 = 3 significant figures so:

1.11M
111k
11.1k
1.1k
1k

Edit: sorry SCI 3, not FIX 3. I don't use this every day

[Csaba Tizedes]

(04-17-2015 09:09 AM)walter b Wrote:  

(04-17-2015 12:33 AM)emece67 Wrote:  ... it also works with units, a capability I really miss in RPN calculators.

Please, would you mind enlightening me why you need that?? TIA.

Background: On our side of the Atlantic Ocean, people use SI units. Doing so, you can throw in any such units (V, A, Ω, F, T, Cb, ...) in an equation and the result will be returned in the corresponding correct unit - as long as the equation is correct, of course.

d:-?

BTW emece67 has right.

I'm mechanical fluid flow and process engineer (we are designs pneumatic conveying systems for power plants). Many times I need to calculate from US/UK units to SI (because our sub-suppliers gives us some data sheet in "imperial" units).

Also in engineering the units has more meaning than they are: a simple example: the ventilator head in pascal is really joule/cubic meter, because 1_Pa = 1_J/m^3.
This means that work what the ventilator gives to the air when the air flows across on the ventilator (this work is drawn as function of flow rate and this is the characteristics curve of fan. Also the ducting has same characteristics curve with meaning of required work for that ducting in selected flow rate. Where the two curve intersect that point where the required work and fan work is equal and the system is operating.). Working with units is really important thing in engineering, not simple "if we calculating everything in SI, all results are SI also".

And of course the material properties or beam section tables: if you calculate 250_kN*m / 45_mm^3, you want to know the result is how many MPa. The unit conversion is very important in everyday work.

Csaba

[walter b]

(04-18-2015 07:48 AM)Csaba Tizedes Wrote:  And of course the material properties or beam section tables: if you calculate 250_kN*m / 45_mm^3, you want to know the result is how many MPa.

It goes without saying that units are essential. Though one should know a good set by heart. Every engineer or scientist should know that Pa = N / m^2 even at midnight. And older engineers and scientists are able to calculate in SCI 1 by heart (or by pencil on a napkin). So 250 / 45 results in some 10% more than 5, and kN*m / mm^3 = 10^3 * 10^9 N m / m^3 = 10^12 Pa, so we end with some 5.5*10^6 MPa or 5.5*10^3 GPa or ...

Keep your math neurons busy else they go on vacation for longer. BTW, that example result "feels" a bit large - did you mean cm^3 instead?

d:-)

Edited to add the napkin.