← Back to Home

Function Generator Specs: Three Numbers, Three Bands

test equipmentfunction generatorspec sheet

The listing says 60 MHz. You put it on a scope to measure the rising edge of a 20 MHz square wave, and the edge comes out sloped, sloped enough that it is no longer a square wave.

I actually read six official manuals from RIGOL, Siglent and UNI-T line by line before this made sense: the machine was fine and my reading of the spec sheet was not. The "frequency" row on that sheet is six rows, not one.

Retail listings print a single number, usually 60 MHz or 120 MHz. In the vendor manual that number has a name: sine. The same table carries separate rows for square, pulse, ramp and arbitrary, and every one of them is lower than the sine row. You buy on the sine row and you work on the others.

This piece re-derives four in-stock models on one consistent basis, and pulls apart the three numbers that get treated as one: frequency, sample rate and amplitude.

TL;DR: three points up front.

Number One: Frequency Only Covers the Sine

The RIGOL DG1000Z manual is the most explicit of the group. It splits frequency into six rows:

Look at the square row. All three machines share one number. Paying for 60 MHz ($529 on the vendor site against $389) buys a wider sine band and nothing at all on square.

Siglent reads the same way. In the SDG2000X manual the sine row is 40, 80 and 120 MHz, while the square row is a single "1μ–25 MHz". So the 120 MHz model also stops its square at 25 MHz, one fourth of its sine figure.

UNI-T spreads wider. Its official selection guide lists the UTG932E at 30 MHz sine, 15 MHz square and 400 kHz ramp. The UTG962E doubles sine to 60 MHz, lifts square to 20 MHz, and leaves the ramp at 400 kHz. The $30 step from $199.99 to $229.00 (prices as scraped on 2026-11-07) buys sine and square, and zero ramp.

That ramp row deserves its own look. At 200 MSa/s, 400 kHz means 500 points per cycle, more than a hundred times the 3.33 points left at 60 MHz sine. So the low ramp ceiling is not a sample rate limit. The vendors do not explain the gap, and from the numbers alone I can confirm exactly one thing: you cannot infer the ramp limit from the sample rate. Two different mechanisms drive those two rows.

Number Two: Sample Rate, Before You Divide by the Interpolation Factor

This is the row people misread most often.

The model table in the Siglent SDG2000X manual reads "Sampling rate 1.2 GSa/s (4X Interpolation)". Most readers skip the parentheses. The same manual, in the arbitrary wave section, gives the real numbers: 1 μSa/s to 75 MSa/s in TrueArb mode, 300 MSa/s in DDS mode.

The official white paper explains the mechanism in plain words: the reference clock of the DDS circuitry on the SDG2000X is 300 MHz, and the interpolation circuit raises the sampling clock of the D/A converter to 1.2 GHz, which is where the 1.2 GSa/s figure with its "4X interpolation" note comes from.

The SDG1000X Plus is written the same way: "1 GSa/s (4X Interpolation)" in the model table, and 250 MSa/s in the arbitrary wave section. One gigahertz divided by four is 250, and the two match.

So a claimed sample rate has to be divided by its interpolation factor before it means anything:

ModelClaimed rateInterpolationReal point clock
RIGOL DG1000Z series200 MSa/snone200 MSa/s
UNI-T UTG900E series200 MSa/snone200 MSa/s
Siglent SDG1000X Plus1 GSa/s4X250 MSa/s
Siglent SDG2000X1.2 GSa/s4X300 MSa/s

Recompute points per cycle at each vendor's own sine ceiling and the listing-page spread largely disappears:

Everything lands in single digits, from 2.5 to 8. Those are the vendors' own sine ceilings, the point where the instrument stops meeting spec, and that leftover headroom belongs to the reconstruction filter, not to your waveform.

Interpolation is not a trick. The white paper is direct about why it exists: a 300 MSa/s D/A can only reach 120 MHz because of reconstruction filter limits, while a 1.2 GSa/s D/A reaches higher, and with 4x interpolation the distortion products of a 120 MHz output sit at 960 MHz and 840 MHz, far outside the passband. What interpolation buys is a cleaner high-frequency sine, not four times the waveform detail.

The arbitrary wave section proves the point: the interpolator pushes the DAC clock to 1.2 GHz, while point-by-point TrueArb output still stops at 75 MSa/s.

Number Three: Amplitude Comes With Two Conditions

Amplitude is the messiest of the three because it carries two conditions at once.

Start with frequency. RIGOL prints the finest table in the group, three tiers into 50Ω:

Walk from 10 MHz to 60 MHz and the usable amplitude falls to one quarter. The 10 Vpp on the listing only exists below 10 MHz.

Siglent gives two tiers and switches the reference to a high-Z load: 2 mV to 20 Vpp up to 20 MHz, 2 mV to 10 Vpp above it. The data sheet also states that 20 Vpp is guaranteed at 20 MHz on both channels, which tells you where the line sits.

UNI-T writes "1 mVpp~10 Vpp (50Ω); 2 mVpp~20 Vpp (high Z)" with no frequency condition at all. That is not unique to UNI-T, but it means the manual cannot tell you whether 10 Vpp survives at 30 MHz. You have to measure it yourself.

Now the load. Siglent adds a footnote under the output characteristics table: "Note 1: The specification will be divided by 2 while applied to a 50Ω load." UNI-T puts the two columns side by side instead: 10 Vpp into 50Ω, 20 Vpp into high Z.

Both statements describe one thing. A 50Ω source impedance and a 50Ω load form a divider, so the load sees half the set value. A scope input is usually 1 MΩ, and there you read twice the set value. Switch to a 50Ω feed-through or a 50Ω terminator and the reading drops back by half. Same instrument, same setting, two readings, 6 dB apart.

One safety number while we are here. The RIGOL manual gives the overvoltage protection thresholds at the output: with amplitude set above 2 Vpp, or offset above 2 VDC, an external voltage beyond ±11.5 V trips protection; with amplitude at or below 2 Vpp, ±3.5 V is enough. The output connector is not an input, so do not clip it onto a live node.

All Three Numbers at Once: What Each Machine Can Actually Do

Put the rows in one table and the models separate:

ModelSineSquarePulseRampArbitrarySquare as % of sine
UNI-T UTG932E30 MHz15 MHz15 MHz400 kHz10 MHz50%
UNI-T UTG962E60 MHz20 MHz20 MHz400 kHz10 MHz33%
RIGOL DG1062Z60 MHz25 MHz25 MHz1 MHz20 MHz42%
Siglent SDG2042X40 MHz25 MHz25 MHz1 MHz20 MHz63%
Siglent SDG2122X120 MHz25 MHz25 MHz1 MHz20 MHz21%

The right column carries the most information. The higher the sine number, the smaller the fraction of it that square gets. The 120 MHz box keeps 21% of its sine, because its square ceiling is identical to the 40 MHz box. The extra 80 MHz belongs to sine.

Four Machines in Stock: Defects Before Prices

Prices and stock below reflect amazon.com pages scraped on 2026-11-07 and move with the page.

UNI-T UTG932E: the entry point that publishes the most rows

30 MHz sine, 15 MHz square, 200 MSa/s, 14 bit, two channels, 50Ω output, amplitude 1 mVpp to 10 Vpp into 50Ω. Scraped at $199.99, page showing only 16 left in stock.

The trade is clear: it publishes what each waveform can do, and charges for it in frequency. The 400 kHz ramp is the weak spot. Fine for sweeping a triangle across the audio band, not enough to sweep into the hundreds of kilohertz.

Real defects: the only interface is USB Device, and the standard accessory list in the official manual has no LAN, no USB Host and no expansion beyond the external trigger input; modulation covers AM, FM, PM and FSK only; the vendor claims 24 built-in arbitrary waveforms but never publishes the arbitrary waveform length; the amplitude row carries no frequency condition, so you cannot learn from the manual how much Vpp survives at 30 MHz.

The product page is here: UNI-T UTG932E.

UNI-T UTG962E: sine doubled, everything else barely moved

60 MHz sine, 20 MHz square, 200 MSa/s, 14 bit, and the same chassis as the 932E at 172 x 90 x 68 mm and 0.33 kg. Scraped at $229.00, page showing in stock.

The $30 buys 2x sine and 1.33x square. At 200 MSa/s and 60 MHz only 3.33 points per cycle remain, one of the lowest figures in the group.

Real defects: the ramp stays at 400 kHz exactly as on the 932E, so the upgrade buys no low-frequency capability at all; 14 bit across a 20 Vpp high-Z range means a 1.22 mV step against 0.305 mV on a 16-bit machine, four times coarser; no LAN, and no published arbitrary waveform length.

The product page is here: UNI-T UTG962E.

SIGLENT SDG2042X: the most balanced of the three numbers

40 MHz sine, 25 MHz square, 1.2 GSa/s (4X interpolation, 300 MSa/s real), 16 bit, 8 Mpts arbitrary, and 20 Vpp up to 20 MHz. Scraped at $574.00, page showing only 2 left in stock.

Its square ceiling matches the 120 MHz model while its sine stops at 40 MHz, which gives it the highest square-to-sine ratio here at 63%. For digital work around 20 MHz it is not far from the two 60 MHz boxes; the real difference is 16 bit and 8 Mpts.

Real defects: the 1.2 GSa/s in the model table is the interpolated DAC clock, the DDS clock is 300 MHz, and point-by-point TrueArb is 75 MSa/s, so a full 1.2 G points does not exist; the ramp tops out at 1 MHz, one fortieth of sine; 20 Vpp holds only below 20 MHz into high Z and halves to 10 Vpp into 50Ω; the page showed 2 units left when scraped.

The product page is here: SIGLENT SDG2042X.

SIGLENT SDG2122X: paying for a 120 MHz sine

120 MHz sine, 25 MHz square, 1.2 GSa/s (4X interpolation), 16 bit, 8 Mpts. Scraped at $1,313.18, page showing in stock.

Its value sits entirely in sine: 120 MHz, with harmonic distortion specified at -38 dBc from 100 to 120 MHz. The square, pulse, ramp and arbitrary rows are identical to the SDG2042X.

Real defects: the 300 MSa/s real clock leaves 2.5 points per cycle at 120 MHz, the lowest here; square stops at 25 MHz, 21% of sine; ramp is 1 MHz, one hundred twentieth of sine; and at 6.6x the price of the UTG932E it buys no higher square, pulse or ramp figure anywhere.

The product page is here: SIGLENT SDG2122X.

One machine used only as a reference

The RIGOL DG1062Z prints the finest amplitude table of the five, with three frequency tiers, but the amazon.com page returned an out-of-stock placeholder when scraped, and the vendor site lists $529 with a note that it ships in 8 weeks. I do not link out-of-stock listings, so it serves only as a reference: 60 MHz sine, 25 MHz square, 200 MSa/s, 14 bit, 8 Mpts, amplitude tiers of 10, 5.0 and 2.5 Vpp.

Pick by What You Need to Measure: Three Branches

If you use one criterion, use the ceiling of the waveform you actually need, not the sine ceiling.

The other side of the same question is when not to buy a bench unit. If you need sines, squares and triangles below 1 MHz, the UTG932E already covers all three rows at 200 MSa/s and 14 bit, and the extra money buys nothing on square or ramp.

Five Minutes of Checks After It Arrives

First, set the frequency to the vendor's ceiling, pick sine, and watch the scope for steps. By the arithmetic above only 2.5 to 8 points per cycle remain up there, so steps are expected; what matters is whether the amplitude holds.

Second, switch to square, set the vendor's square ceiling, and look at the rising edge. The published edge time is a typical value at 1 Vpp into 50Ω. Set 10 Vpp and the edge slows down, which the RIGOL manual states in the condition column rather than the body text.

Third, change the load from 1 MΩ to 50Ω and watch the voltage. Halving is correct. No change means your 50Ω termination is not seated.

Fourth, wait 30 minutes before judging accuracy. Both RIGOL and Siglent require the same conditions: within the calibration period, continuously powered for at least 30 minutes, at 18 to 28 degrees Celsius. A number taken at power-on does not count.

The Boundaries of This Piece

I did not bench-test all four units. Every frequency, amplitude, edge and load condition here comes from vendor manual text: the RIGOL DG1000Z data sheet DSB09103-1110, the Siglent SDG2000X data sheet EN02H, the Siglent SDG1000X Plus data sheet EN01A, and the UNI-T UTG900E series selection guide, all retrieved on 2026-11-07. Prices and stock are amazon.com page state from the same day. Anything derived, such as points per cycle, the edge share of a period or the output current after the 50Ω halving, is shown with its arithmetic so you can recompute it. Why the ramp ceiling sits so far below the sine ceiling is not explained in any of the three manuals, so it stays marked as unexplained rather than turned into a conclusion.

FAQ

Q: Can a 60 MHz function generator output a 60 MHz square wave?

No. Square is its own row in the manual: 20 MHz on the UTG962E, 25 MHz on the RIGOL DG1062Z, 25 MHz on the SDG2122X. The 60 MHz or 120 MHz figure applies to sine alone.

Q: Is 1.2 GSa/s six times better than 200 MSa/s?

No. Siglent prints "1.2 GSa/s (4X Interpolation)", and its white paper states the DDS clock is 300 MHz with the interpolator raising the DAC clock to 1.2 GHz. Divided by four that is 300 MSa/s against 200 MSa/s, a factor of 1.5. Interpolation moves the distortion products of a high-frequency sine further out; it does not add waveform detail.

Q: Why does 20 Vpp read as 10 Vpp on my scope?

A 50Ω source impedance and a 50Ω load form a divider. Siglent states it in Note 1: "The specification will be divided by 2 while applied to a 50Ω load."

Q: How much does 14 bit versus 16 bit matter?

Across a 20 Vpp range, 14 bit gives 16,384 levels and roughly a 1.22 mV step; 16 bit gives 65,536 levels and roughly 0.305 mV. That is a fourfold difference in step height. It shows up when you stack a small signal on a large one or test ADC linearity, and it mostly disappears when you drive digital timing.

Q: Do I need 8 Mpts of memory?

Depends on playback length. 8 Mpts is 40 ms at 200 MSa/s and 26.7 ms at 300 MSa/s. Beyond that you loop or use segmented sequences. For standard waveforms the memory depth does nothing for you.

Q: Why is the ramp ceiling so low?

The vendors do not say. What can be confirmed is the arithmetic: ramp runs from 400 kHz to 2 MHz while sine runs from 25 MHz to 120 MHz, and 400 kHz at 200 MSa/s leaves 500 points per cycle, far more than the 3.33 points at 60 MHz sine. So this is not a sample rate limit. Buy against the ramp row, and do not derive it from the sine row.

About the Links

This article contains Amazon affiliate links. If you order through them the site may earn a commission, your price does not change, and it does not affect the judgments above. Every number is sourced: frequency, amplitude, edge and load conditions come from the RIGOL DG1000Z data sheet, the Siglent SDG2000X and SDG1000X Plus data sheets, Siglent's official interpolated DDS white paper, and the UNI-T UTG900E series selection guide. Prices and stock reflect amazon.com pages scraped on 2026-11-07.

If your workflow starts with the waveform and not the signal, the 100 MHz oscilloscope breakdown applies the same question to channel count and sample rate; for timing capture rather than waveform viewing, the logic analyzer channel and sample rate math is the same arithmetic; and the DC electronic load piece takes 150V, 30A and 200W apart the same way.

Back to that 60 MHz box: it was never wrong. Treating the sine row as the whole table was. What decides whether you can measure anything is the 20 MHz on the square row and the 10 ns in the edge column.

👉 Join Xiaomi MiMo Platform: Leading AI model platform with cost-effective inference

👉 Join Aliyun AI: Top AI products with exclusive coupons for business innovation

📌 This article was AI-assisted generated and human-reviewed | TechPassive — An AI-driven content testing site focused on real tool reviews

🔗 Recommended Tools

These are carefully selected tools. Using our affiliate links supports us to keep producing quality content:

As an Amazon Associate, I earn from qualifying purchases at no extra cost to you.

☁️ DigitalOcean Cloud ⚡ Vultr VPS 🤖 QoderWork CN (Refer & Earn) ☁️ Aliyun AI Products 📚 WordPress Books 🧠 AI & LLM Books 🏗️ System Design Books 🐳 Docker & K8s Books 🐧 Linux Admin Books 🐍 Python Books 🗄️ Database Books 🚀 DevOps & SRE Books 🤖 Xiaomi MiMo Platform
← Back to Home